chore(deps-dev): bump pre-commit from 4.5.0 to 4.5.1

Bumps [pre-commit](https://github.com/pre-commit/pre-commit) from 4.5.0 to 4.5.1.
- [Release notes](https://github.com/pre-commit/pre-commit/releases)
- [Changelog](https://github.com/pre-commit/pre-commit/blob/main/CHANGELOG.md)
- [Commits](https://github.com/pre-commit/pre-commit/compare/v4.5.0...v4.5.1)

---
updated-dependencies:
- dependency-name: pre-commit
  dependency-version: 4.5.1
  dependency-type: direct:development
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>

.github/dependabot.yml

@@ -5,7 +5,12 @@
 
 version: 2
 updates:
-  - package-ecosystem: uv # See documentation for possible values
-    directory: "/" # Location of package manifests
+  - package-ecosystem: uv
+    directory: "/"
     schedule:
       interval: "weekly"
+    groups:
+      security-updates:
+        applies-to: security-updates
+        patterns:
+          - "*"

.github/workflows/generate-tool-specs.yml

@@ -0,0 +1,63 @@
+name: Generate Tool Specifications
+
+on:
+  pull_request:
+    branches:
+      - main
+    paths:
+      - 'lib/crewai-tools/src/crewai_tools/**'
+  workflow_dispatch:
+
+permissions:
+  contents: write
+  pull-requests: write
+
+jobs:
+  generate-specs:
+    runs-on: ubuntu-latest
+    env:
+      PYTHONUNBUFFERED: 1
+
+    steps:
+      - name: Generate GitHub App token
+        id: app-token
+        uses: tibdex/github-app-token@v2
+        with:
+          app_id: ${{ secrets.CREWAI_TOOL_SPECS_APP_ID }}
+          private_key: ${{ secrets.CREWAI_TOOL_SPECS_PRIVATE_KEY }}
+
+      - name: Checkout code
+        uses: actions/checkout@v4
+        with:
+          ref: ${{ github.head_ref }}
+          token: ${{ steps.app-token.outputs.token }}
+
+      - name: Install uv
+        uses: astral-sh/setup-uv@v6
+        with:
+          version: "0.8.4"
+          python-version: "3.12"
+          enable-cache: true
+
+      - name: Install the project
+        working-directory: lib/crewai-tools
+        run: uv sync --dev --all-extras
+
+      - name: Generate tool specifications
+        working-directory: lib/crewai-tools
+        run: uv run python src/crewai_tools/generate_tool_specs.py
+
+      - name: Check for changes and commit
+        run: |
+          git config user.name "github-actions[bot]"
+          git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
+
+          git add lib/crewai-tools/tool.specs.json
+
+          if git diff --quiet --staged; then
+            echo "No changes detected in tool.specs.json"
+          else
+            echo "Changes detected in tool.specs.json, committing..."
+            git commit -m "chore: update tool specifications"
+            git push
+          fi

.python-version

@@ -0,0 +1 @@
+3.13

lib/crewai-tools/src/crewai_tools/adapters/mcp_adapter.py

@@ -2,29 +2,95 @@
 
 from __future__ import annotations
 
+from collections.abc import Callable
 import logging
 from typing import TYPE_CHECKING, Any
 
 from crewai.tools import BaseTool
+from crewai.utilities.pydantic_schema_utils import create_model_from_schema
+from crewai.utilities.string_utils import sanitize_tool_name
+from pydantic import BaseModel
 
 from crewai_tools.adapters.tool_collection import ToolCollection
 
 
-logger = logging.getLogger(__name__)
-
 if TYPE_CHECKING:
     from mcp import StdioServerParameters
-    from mcpadapt.core import MCPAdapt
-    from mcpadapt.crewai_adapter import CrewAIAdapter
+    from mcp.types import CallToolResult, TextContent, Tool
+    from mcpadapt.core import MCPAdapt, ToolAdapter
+
+
+logger = logging.getLogger(__name__)
 
 
 try:
     from mcp import StdioServerParameters
-    from mcpadapt.core import MCPAdapt
-    from mcpadapt.crewai_adapter import CrewAIAdapter
+    from mcp.types import CallToolResult, TextContent, Tool
+    from mcpadapt.core import MCPAdapt, ToolAdapter
+
+    class CrewAIToolAdapter(ToolAdapter):
+        """Adapter that creates CrewAI tools with properly normalized JSON schemas.
+
+        This adapter bypasses mcpadapt's model creation which adds invalid null values
+        to field schemas, instead using CrewAI's own schema utilities.
+        """
+
+        def adapt(
+            self,
+            func: Callable[[dict[str, Any] | None], CallToolResult],
+            mcp_tool: Tool,
+        ) -> BaseTool:
+            """Adapt a MCP tool to a CrewAI tool.
+
+            Args:
+                func: The function to call when the tool is invoked.
+                mcp_tool: The MCP tool definition to adapt.
+
+            Returns:
+                A CrewAI BaseTool instance.
+            """
+            tool_name = sanitize_tool_name(mcp_tool.name)
+            tool_description = mcp_tool.description or ""
+            args_model = create_model_from_schema(mcp_tool.inputSchema)
+
+            class CrewAIMCPTool(BaseTool):
+                name: str = tool_name
+                description: str = tool_description
+                args_schema: type[BaseModel] = args_model
+
+                def _run(self, **kwargs: Any) -> Any:
+                    result = func(kwargs)
+                    if len(result.content) == 1:
+                        first_content = result.content[0]
+                        if isinstance(first_content, TextContent):
+                            return first_content.text
+                        return str(first_content)
+                    return str(
+                        [
+                            content.text
+                            for content in result.content
+                            if isinstance(content, TextContent)
+                        ]
+                    )
+
+                def _generate_description(self) -> None:
+                    schema = self.args_schema.model_json_schema()
+                    schema.pop("$defs", None)
+                    self.description = (
+                        f"Tool Name: {self.name}\n"
+                        f"Tool Arguments: {schema}\n"
+                        f"Tool Description: {self.description}"
+                    )
+
+            return CrewAIMCPTool()
+
+        async def async_adapt(self, afunc: Any, mcp_tool: Tool) -> Any:
+            """Async adaptation is not supported by CrewAI."""
+            raise NotImplementedError("async is not supported by the CrewAI framework.")
 
     MCP_AVAILABLE = True
-except ImportError:
+except ImportError as e:
+    logger.debug(f"MCP packages not available: {e}")
     MCP_AVAILABLE = False
 
 
@@ -34,9 +100,6 @@ class MCPServerAdapter:
     Note: tools can only be accessed after the server has been started with the
         `start()` method.
 
-    Attributes:
-        tools: The CrewAI tools available from the MCP server.
-
     Usage:
         # context manager + stdio
         with MCPServerAdapter(...) as tools:
@@ -89,7 +152,9 @@ class MCPServerAdapter:
         super().__init__()
         self._adapter = None
         self._tools = None
-        self._tool_names = list(tool_names) if tool_names else None
+        self._tool_names = (
+            [sanitize_tool_name(name) for name in tool_names] if tool_names else None
+        )
 
         if not MCP_AVAILABLE:
             import click
@@ -100,7 +165,7 @@ class MCPServerAdapter:
                 import subprocess
 
                 try:
-                    subprocess.run(["uv", "add", "mcp crewai-tools[mcp]"], check=True)  # noqa: S607
+                    subprocess.run(["uv", "add", "mcp crewai-tools'[mcp]'"], check=True)  # noqa: S607
 
                 except subprocess.CalledProcessError as e:
                     raise ImportError("Failed to install mcp package") from e
@@ -112,7 +177,7 @@ class MCPServerAdapter:
         try:
             self._serverparams = serverparams
             self._adapter = MCPAdapt(
-                self._serverparams, CrewAIAdapter(), connect_timeout
+                self._serverparams, CrewAIToolAdapter(), connect_timeout
             )
             self.start()
 
@@ -124,13 +189,13 @@ class MCPServerAdapter:
                     logger.error(f"Error during stop cleanup: {stop_e}")
             raise RuntimeError(f"Failed to initialize MCP Adapter: {e}") from e
 
-    def start(self):
+    def start(self) -> None:
         """Start the MCP server and initialize the tools."""
-        self._tools = self._adapter.__enter__()
+        self._tools = self._adapter.__enter__()  # type: ignore[union-attr]
 
-    def stop(self):
+    def stop(self) -> None:
         """Stop the MCP server."""
-        self._adapter.__exit__(None, None, None)
+        self._adapter.__exit__(None, None, None)  # type: ignore[union-attr]
 
     @property
     def tools(self) -> ToolCollection[BaseTool]:
@@ -152,12 +217,19 @@ class MCPServerAdapter:
             return tools_collection.filter_by_names(self._tool_names)
         return tools_collection
 
-    def __enter__(self):
-        """Enter the context manager. Note that `__init__()` already starts the MCP server.
-        So tools should already be available.
+    def __enter__(self) -> ToolCollection[BaseTool]:
+        """Enter the context manager.
+
+        Note that `__init__()` already starts the MCP server,
+        so tools should already be available.
         """
         return self.tools
 
-    def __exit__(self, exc_type, exc_value, traceback):
+    def __exit__(
+        self,
+        exc_type: type[BaseException] | None,
+        exc_value: BaseException | None,
+        traceback: Any,
+    ) -> None:
         """Exit the context manager."""
-        return self._adapter.__exit__(exc_type, exc_value, traceback)
+        self._adapter.__exit__(exc_type, exc_value, traceback)  # type: ignore[union-attr]

lib/crewai-tools/src/crewai_tools/generate_tool_specs.py

@@ -8,8 +8,9 @@ from typing import Any
 
 from crewai.tools.base_tool import BaseTool, EnvVar
 from pydantic import BaseModel
+from pydantic.fields import FieldInfo
 from pydantic.json_schema import GenerateJsonSchema
-from pydantic_core import PydanticOmit
+from pydantic_core import PydanticOmit, PydanticUndefined
 
 from crewai_tools import tools
 
@@ -44,6 +45,9 @@ class ToolSpecExtractor:
             schema = self._unwrap_schema(core_schema)
             fields = schema.get("schema", {}).get("fields", {})
 
+            # Use model_fields to get defaults (handles both default and default_factory)
+            model_fields = tool_class.model_fields
+
             tool_info = {
                 "name": tool_class.__name__,
                 "humanized_name": self._extract_field_default(
@@ -54,9 +58,9 @@ class ToolSpecExtractor:
                 ).strip(),
                 "run_params_schema": self._extract_params(fields.get("args_schema")),
                 "init_params_schema": self._extract_init_params(tool_class),
-                "env_vars": self._extract_env_vars(fields.get("env_vars")),
-                "package_dependencies": self._extract_field_default(
-                    fields.get("package_dependencies"), fallback=[]
+                "env_vars": self._extract_env_vars_from_model_fields(model_fields),
+                "package_dependencies": self._extract_package_deps_from_model_fields(
+                    model_fields
                 ),
             }
 
@@ -103,10 +107,27 @@ class ToolSpecExtractor:
             return {}
 
     @staticmethod
-    def _extract_env_vars(
-        env_vars_field: dict[str, Any] | None,
+    def _get_field_default(field: FieldInfo | None) -> Any:
+        """Get default value from a FieldInfo, handling both default and default_factory."""
+        if not field:
+            return None
+
+        default_value = field.default
+        if default_value is PydanticUndefined or default_value is None:
+            if field.default_factory:
+                return field.default_factory()
+            return None
+
+        return default_value
+
+    @staticmethod
+    def _extract_env_vars_from_model_fields(
+        model_fields: dict[str, FieldInfo],
     ) -> list[dict[str, Any]]:
-        if not env_vars_field:
+        default_value = ToolSpecExtractor._get_field_default(
+            model_fields.get("env_vars")
+        )
+        if not default_value:
             return []
 
         return [
@@ -116,10 +137,22 @@ class ToolSpecExtractor:
                 "required": env_var.required,
                 "default": env_var.default,
             }
-            for env_var in env_vars_field.get("schema", {}).get("default", [])
+            for env_var in default_value
             if isinstance(env_var, EnvVar)
         ]
 
+    @staticmethod
+    def _extract_package_deps_from_model_fields(
+        model_fields: dict[str, FieldInfo],
+    ) -> list[str]:
+        default_value = ToolSpecExtractor._get_field_default(
+            model_fields.get("package_dependencies")
+        )
+        if not isinstance(default_value, list):
+            return []
+
+        return default_value
+
     @staticmethod
     def _extract_init_params(tool_class: type[BaseTool]) -> dict[str, Any]:
         ignored_init_params = [
@@ -152,7 +185,7 @@ class ToolSpecExtractor:
 
 
 if __name__ == "__main__":
-    output_file = Path(__file__).parent / "tool.specs.json"
+    output_file = Path(__file__).parent.parent.parent / "tool.specs.json"
     extractor = ToolSpecExtractor()
 
     extractor.extract_all_tools()

lib/crewai-tools/src/crewai_tools/tools/brave_search_tool/brave_search_tool.py

@@ -1,12 +1,17 @@
 from datetime import datetime
+import json
 import os
 import time
-from typing import Any, ClassVar
+from typing import Annotated, Any, ClassVar, Literal
 
 from crewai.tools import BaseTool, EnvVar
+from dotenv import load_dotenv
 from pydantic import BaseModel, Field
+from pydantic.types import StringConstraints
 import requests
 
+load_dotenv()
+
 
 def _save_results_to_file(content: str) -> None:
     """Saves the search results to a file."""
@@ -15,37 +20,72 @@ def _save_results_to_file(content: str) -> None:
         file.write(content)
 
 
-class BraveSearchToolSchema(BaseModel):
-    """Input for BraveSearchTool."""
+FreshnessPreset = Literal["pd", "pw", "pm", "py"]
+FreshnessRange = Annotated[
+    str, StringConstraints(pattern=r"^\d{4}-\d{2}-\d{2}to\d{4}-\d{2}-\d{2}$")
+]
+Freshness = FreshnessPreset | FreshnessRange
+SafeSearch = Literal["off", "moderate", "strict"]
 
-    search_query: str = Field(
-        ..., description="Mandatory search query you want to use to search the internet"
+
+class BraveSearchToolSchema(BaseModel):
+    """Input for BraveSearchTool"""
+
+    query: str = Field(..., description="Search query to perform")
+    country: str | None = Field(
+        default=None,
+        description="Country code for geo-targeting (e.g., 'US', 'BR').",
+    )
+    search_language: str | None = Field(
+        default=None,
+        description="Language code for the search results (e.g., 'en', 'es').",
+    )
+    count: int | None = Field(
+        default=None,
+        description="The maximum number of results to return. Actual number may be less.",
+    )
+    offset: int | None = Field(
+        default=None, description="Skip the first N result sets/pages. Max is 9."
+    )
+    safesearch: SafeSearch | None = Field(
+        default=None,
+        description="Filter out explicit content. Options: off/moderate/strict",
+    )
+    spellcheck: bool | None = Field(
+        default=None,
+        description="Attempt to correct spelling errors in the search query.",
+    )
+    freshness: Freshness | None = Field(
+        default=None,
+        description="Enforce freshness of results. Options: pd/pw/pm/py, or YYYY-MM-DDtoYYYY-MM-DD",
+    )
+    text_decorations: bool | None = Field(
+        default=None,
+        description="Include markup to highlight search terms in the results.",
+    )
+    extra_snippets: bool | None = Field(
+        default=None,
+        description="Include up to 5 text snippets for each page if possible.",
+    )
+    operators: bool | None = Field(
+        default=None,
+        description="Whether to apply search operators (e.g., site:example.com).",
     )
 
 
+# TODO: Extend support to additional endpoints (e.g., /images, /news, etc.)
 class BraveSearchTool(BaseTool):
-    """BraveSearchTool - A tool for performing web searches using the Brave Search API.
+    """A tool that performs web searches using the Brave Search API."""
 
-    This module provides functionality to search the internet using Brave's Search API,
-    supporting customizable result counts and country-specific searches.
-
-    Dependencies:
-        - requests
-        - pydantic
-        - python-dotenv (for API key management)
-    """
-
-    name: str = "Brave Web Search the internet"
+    name: str = "Brave Search"
     description: str = (
-        "A tool that can be used to search the internet with a search_query."
+        "A tool that performs web searches using the Brave Search API. "
+        "Results are returned as structured JSON data."
     )
     args_schema: type[BaseModel] = BraveSearchToolSchema
     search_url: str = "https://api.search.brave.com/res/v1/web/search"
-    country: str | None = ""
     n_results: int = 10
     save_file: bool = False
-    _last_request_time: ClassVar[float] = 0
-    _min_request_interval: ClassVar[float] = 1.0  # seconds
     env_vars: list[EnvVar] = Field(
         default_factory=lambda: [
             EnvVar(
@@ -55,6 +95,9 @@ class BraveSearchTool(BaseTool):
             ),
         ]
     )
+    # Rate limiting parameters
+    _last_request_time: ClassVar[float] = 0
+    _min_request_interval: ClassVar[float] = 1.0  # seconds
 
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
@@ -73,19 +116,64 @@ class BraveSearchTool(BaseTool):
                 self._min_request_interval - (current_time - self._last_request_time)
             )
         BraveSearchTool._last_request_time = time.time()
+
+        # Construct and send the request
         try:
-            search_query = kwargs.get("search_query") or kwargs.get("query")
-            if not search_query:
-                raise ValueError("Search query is required")
+            # Maintain both "search_query" and "query" for backwards compatibility
+            query = kwargs.get("search_query") or kwargs.get("query")
+            if not query:
+                raise ValueError("Query is required")
+
+            payload = {"q": query}
+
+            if country := kwargs.get("country"):
+                payload["country"] = country
+
+            if search_language := kwargs.get("search_language"):
+                payload["search_language"] = search_language
+
+            # Fallback to deprecated n_results parameter if no count is provided
+            count = kwargs.get("count")
+            if count is not None:
+                payload["count"] = count
+            else:
+                payload["count"] = self.n_results
+
+            # Offset may be 0, so avoid truthiness check
+            offset = kwargs.get("offset")
+            if offset is not None:
+                payload["offset"] = offset
+
+            if safesearch := kwargs.get("safesearch"):
+                payload["safesearch"] = safesearch
 
             save_file = kwargs.get("save_file", self.save_file)
-            n_results = kwargs.get("n_results", self.n_results)
+            if freshness := kwargs.get("freshness"):
+                payload["freshness"] = freshness
 
-            payload = {"q": search_query, "count": n_results}
+            # Boolean parameters
+            spellcheck = kwargs.get("spellcheck")
+            if spellcheck is not None:
+                payload["spellcheck"] = spellcheck
 
-            if self.country != "":
-                payload["country"] = self.country
+            text_decorations = kwargs.get("text_decorations")
+            if text_decorations is not None:
+                payload["text_decorations"] = text_decorations
 
+            extra_snippets = kwargs.get("extra_snippets")
+            if extra_snippets is not None:
+                payload["extra_snippets"] = extra_snippets
+
+            operators = kwargs.get("operators")
+            if operators is not None:
+                payload["operators"] = operators
+
+            # Limit the result types to "web" since there is presently no
+            # handling of other types like "discussions", "faq", "infobox",
+            # "news", "videos", or "locations".
+            payload["result_filter"] = "web"
+
+            # Setup Request Headers
             headers = {
                 "X-Subscription-Token": os.environ["BRAVE_API_KEY"],
                 "Accept": "application/json",
@@ -97,25 +185,32 @@ class BraveSearchTool(BaseTool):
             response.raise_for_status()  # Handle non-200 responses
             results = response.json()
 
+            # TODO: Handle other result types like "discussions", "faq", etc.
+            web_results_items = []
             if "web" in results:
-                results = results["web"]["results"]
-                string = []
-                for result in results:
-                    try:
-                        string.append(
-                            "\n".join(
-                                [
-                                    f"Title: {result['title']}",
-                                    f"Link: {result['url']}",
-                                    f"Snippet: {result['description']}",
-                                    "---",
-                                ]
-                            )
-                        )
-                    except KeyError:  # noqa: PERF203
-                        continue
+                web_results = results["web"]["results"]
 
-            content = "\n".join(string)
+                for result in web_results:
+                    url = result.get("url")
+                    title = result.get("title")
+                    # If, for whatever reason, this entry does not have a title
+                    # or url, skip it.
+                    if not url or not title:
+                        continue
+                    item = {
+                        "url": url,
+                        "title": title,
+                    }
+                    description = result.get("description")
+                    if description:
+                        item["description"] = description
+                    snippets = result.get("extra_snippets")
+                    if snippets:
+                        item["snippets"] = snippets
+
+                    web_results_items.append(item)
+
+            content = json.dumps(web_results_items)
         except requests.RequestException as e:
             return f"Error performing search: {e!s}"
         except KeyError as e:

lib/crewai-tools/src/crewai_tools/tools/stagehand_tool/stagehand_tool.py

@@ -4,7 +4,7 @@ import os
 import re
 from typing import Any
 
-from crewai.tools import BaseTool
+from crewai.tools import BaseTool, EnvVar
 from pydantic import BaseModel, Field
 
 
@@ -137,6 +137,21 @@ class StagehandTool(BaseTool):
     - 'observe': For finding elements in a specific area
     """
     args_schema: type[BaseModel] = StagehandToolSchema
+    package_dependencies: list[str] = Field(default_factory=lambda: ["stagehand<=0.5.9"])
+    env_vars: list[EnvVar] = Field(
+        default_factory=lambda: [
+            EnvVar(
+                name="BROWSERBASE_API_KEY",
+                description="API key for Browserbase services",
+                required=False,
+            ),
+            EnvVar(
+                name="BROWSERBASE_PROJECT_ID",
+                description="Project ID for Browserbase services",
+                required=False,
+            ),
+        ]
+    )
 
     # Stagehand configuration
     api_key: str | None = None

lib/crewai-tools/tests/test_generate_tool_specs.py

@@ -23,23 +23,26 @@ class MockTool(BaseTool):
     )
     my_parameter: str = Field("This is default value", description="What a description")
     my_parameter_bool: bool = Field(False)
+    # Use default_factory like real tools do (not direct default)
     package_dependencies: list[str] = Field(
-        ["this-is-a-required-package", "another-required-package"], description=""
+        default_factory=lambda: ["this-is-a-required-package", "another-required-package"]
+    )
+    env_vars: list[EnvVar] = Field(
+        default_factory=lambda: [
+            EnvVar(
+                name="SERPER_API_KEY",
+                description="API key for Serper",
+                required=True,
+                default=None,
+            ),
+            EnvVar(
+                name="API_RATE_LIMIT",
+                description="API rate limit",
+                required=False,
+                default="100",
+            ),
+        ]
     )
-    env_vars: list[EnvVar] = [
-        EnvVar(
-            name="SERPER_API_KEY",
-            description="API key for Serper",
-            required=True,
-            default=None,
-        ),
-        EnvVar(
-            name="API_RATE_LIMIT",
-            description="API rate limit",
-            required=False,
-            default="100",
-        ),
-    ]
 
 
 @pytest.fixture

lib/crewai-tools/tests/tools/brave_search_tool_test.py

@@ -1,8 +1,10 @@
+import json
 from unittest.mock import patch
 
-from crewai_tools.tools.brave_search_tool.brave_search_tool import BraveSearchTool
 import pytest
 
+from crewai_tools.tools.brave_search_tool.brave_search_tool import BraveSearchTool
+
 
 @pytest.fixture
 def brave_tool():
@@ -30,16 +32,43 @@ def test_brave_tool_search(mock_get, brave_tool):
     }
     mock_get.return_value.json.return_value = mock_response
 
-    result = brave_tool.run(search_query="test")
+    result = brave_tool.run(query="test")
     assert "Test Title" in result
     assert "http://test.com" in result
 
 
-def test_brave_tool():
-    tool = BraveSearchTool(
-        n_results=2,
-    )
-    tool.run(search_query="ChatGPT")
+@patch("requests.get")
+def test_brave_tool(mock_get):
+    mock_response = {
+        "web": {
+            "results": [
+                {
+                    "title": "Brave Browser",
+                    "url": "https://brave.com",
+                    "description": "Brave Browser description",
+                }
+            ]
+        }
+    }
+    mock_get.return_value.json.return_value = mock_response
+
+    tool = BraveSearchTool(n_results=2)
+    result = tool.run(query="Brave Browser")
+    assert result is not None
+
+    # Parse JSON so we can examine the structure
+    data = json.loads(result)
+    assert isinstance(data, list)
+    assert len(data) >= 1
+
+    # First item should have expected fields: title, url, and description
+    first = data[0]
+    assert "title" in first
+    assert first["title"] == "Brave Browser"
+    assert "url" in first
+    assert first["url"] == "https://brave.com"
+    assert "description" in first
+    assert first["description"] == "Brave Browser description"
 
 
 if __name__ == "__main__":

lib/crewai/pyproject.toml

@@ -10,7 +10,7 @@ requires-python = ">=3.10, <3.14"
 dependencies = [
     # Core Dependencies
     "pydantic~=2.11.9",
-    "openai~=1.83.0",
+    "openai>=1.83.0,<3",
     "instructor>=1.3.3",
     # Text Processing
     "pdfplumber~=0.11.4",
@@ -78,7 +78,7 @@ voyageai = [
     "voyageai~=0.3.5",
 ]
 litellm = [
-    "litellm~=1.74.9",
+    "litellm>=1.74.9,<3",
 ]
 bedrock = [
     "boto3~=1.40.45",

lib/crewai/src/crewai/agent/core.py

@@ -118,6 +118,8 @@ MCP_TOOL_EXECUTION_TIMEOUT: Final[int] = 30
 MCP_DISCOVERY_TIMEOUT: Final[int] = 15
 MCP_MAX_RETRIES: Final[int] = 3
 
+_passthrough_exceptions: tuple[type[Exception], ...] = ()
+
 # Simple in-memory cache for MCP tool schemas (duration: 5 minutes)
 _mcp_schema_cache: dict[str, Any] = {}
 _cache_ttl: Final[int] = 300  # 5 minutes
@@ -479,6 +481,8 @@ class Agent(BaseAgent):
                     ),
                 )
                 raise e
+            if isinstance(e, _passthrough_exceptions):
+                raise
             self._times_executed += 1
             if self._times_executed > self.max_retry_limit:
                 crewai_event_bus.emit(
@@ -711,6 +715,8 @@ class Agent(BaseAgent):
                     ),
                 )
                 raise e
+            if isinstance(e, _passthrough_exceptions):
+                raise
             self._times_executed += 1
             if self._times_executed > self.max_retry_limit:
                 crewai_event_bus.emit(

lib/crewai/src/crewai/agents/agent_adapters/base_agent_adapter.py

@@ -37,9 +37,10 @@ class BaseAgentAdapter(BaseAgent, ABC):
             tools: Optional list of BaseTool instances to be configured
         """
 
-    def configure_structured_output(self, structured_output: Any) -> None:
+    @abstractmethod
+    def configure_structured_output(self, task: Any) -> None:
         """Configure the structured output for the specific agent implementation.
 
         Args:
-            structured_output: The structured output to be configured
+            task: The task object containing output format specifications.
         """

lib/crewai/src/crewai/agents/agent_adapters/openai_agents/openai_agent_tool_adapter.py

@@ -16,6 +16,7 @@ from crewai.agents.agent_adapters.openai_agents.protocols import (
 )
 from crewai.tools import BaseTool
 from crewai.utilities.import_utils import require
+from crewai.utilities.pydantic_schema_utils import force_additional_properties_false
 from crewai.utilities.string_utils import sanitize_tool_name
 
 
@@ -135,7 +136,9 @@ class OpenAIAgentToolAdapter(BaseToolAdapter):
         for tool in tools:
             schema: dict[str, Any] = tool.args_schema.model_json_schema()
 
-            schema.update({"additionalProperties": False, "type": "object"})
+            schema = force_additional_properties_false(schema)
+
+            schema.update({"type": "object"})
 
             openai_tool: OpenAIFunctionTool = cast(
                 OpenAIFunctionTool,

lib/crewai/src/crewai/agents/agent_builder/base_agent_executor_mixin.py

@@ -4,7 +4,6 @@ import time
 from typing import TYPE_CHECKING
 
 from crewai.agents.parser import AgentFinish
-from crewai.events.event_listener import event_listener
 from crewai.memory.entity.entity_memory_item import EntityMemoryItem
 from crewai.memory.long_term.long_term_memory_item import LongTermMemoryItem
 from crewai.utilities.converter import ConverterError
@@ -138,52 +137,3 @@ class CrewAgentExecutorMixin:
                     content="Long term memory is enabled, but entity memory is not enabled. Please configure entity memory or set memory=True to automatically enable it.",
                     color="bold_yellow",
                 )
-
-    def _ask_human_input(self, final_answer: str) -> str:
-        """Prompt human input with mode-appropriate messaging.
-
-        Note: The final answer is already displayed via the AgentLogsExecutionEvent
-        panel, so we only show the feedback prompt here.
-        """
-        from rich.panel import Panel
-        from rich.text import Text
-
-        formatter = event_listener.formatter
-        formatter.pause_live_updates()
-
-        try:
-            # Training mode prompt (single iteration)
-            if self.crew and getattr(self.crew, "_train", False):
-                prompt_text = (
-                    "TRAINING MODE: Provide feedback to improve the agent's performance.\n\n"
-                    "This will be used to train better versions of the agent.\n"
-                    "Please provide detailed feedback about the result quality and reasoning process."
-                )
-                title = "🎓 Training Feedback Required"
-            # Regular human-in-the-loop prompt (multiple iterations)
-            else:
-                prompt_text = (
-                    "Provide feedback on the Final Result above.\n\n"
-                    "• If you are happy with the result, simply hit Enter without typing anything.\n"
-                    "• Otherwise, provide specific improvement requests.\n"
-                    "• You can provide multiple rounds of feedback until satisfied."
-                )
-                title = "💬 Human Feedback Required"
-
-            content = Text()
-            content.append(prompt_text, style="yellow")
-
-            prompt_panel = Panel(
-                content,
-                title=title,
-                border_style="yellow",
-                padding=(1, 2),
-            )
-            formatter.console.print(prompt_panel)
-
-            response = input()
-            if response.strip() != "":
-                formatter.console.print("\n[cyan]Processing your feedback...[/cyan]")
-            return response
-        finally:
-            formatter.resume_live_updates()

lib/crewai/src/crewai/agents/crew_agent_executor.py

@@ -19,6 +19,7 @@ from crewai.agents.parser import (
     AgentFinish,
     OutputParserError,
 )
+from crewai.core.providers.human_input import ExecutorContext, get_provider
 from crewai.events.event_bus import crewai_event_bus
 from crewai.events.types.logging_events import (
     AgentLogsExecutionEvent,
@@ -175,15 +176,16 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
         """
         return self.llm.supports_stop_words() if self.llm else False
 
-    def invoke(self, inputs: dict[str, Any]) -> dict[str, Any]:
-        """Execute the agent with given inputs.
+    def _setup_messages(self, inputs: dict[str, Any]) -> None:
+        """Set up messages for the agent execution.
 
         Args:
             inputs: Input dictionary containing prompt variables.
-
-        Returns:
-            Dictionary with agent output.
         """
+        provider = get_provider()
+        if provider.setup_messages(cast(ExecutorContext, cast(object, self))):
+            return
+
         if "system" in self.prompt:
             system_prompt = self._format_prompt(
                 cast(str, self.prompt.get("system", "")), inputs
@@ -197,6 +199,19 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
             user_prompt = self._format_prompt(self.prompt.get("prompt", ""), inputs)
             self.messages.append(format_message_for_llm(user_prompt))
 
+        provider.post_setup_messages(cast(ExecutorContext, cast(object, self)))
+
+    def invoke(self, inputs: dict[str, Any]) -> dict[str, Any]:
+        """Execute the agent with given inputs.
+
+        Args:
+            inputs: Input dictionary containing prompt variables.
+
+        Returns:
+            Dictionary with agent output.
+        """
+        self._setup_messages(inputs)
+
         self._inject_multimodal_files(inputs)
 
         self._show_start_logs()
@@ -799,6 +814,7 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                 agent_key=agent_key,
             ),
         )
+        error_event_emitted = False
 
         track_delegation_if_needed(func_name, args_dict, self.task)
 
@@ -881,6 +897,7 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                             error=e,
                         ),
                     )
+                    error_event_emitted = True
         elif max_usage_reached and original_tool:
             # Return error message when max usage limit is reached
             result = f"Tool '{func_name}' has reached its usage limit of {original_tool.max_usage_count} times and cannot be used anymore."
@@ -908,20 +925,20 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                     color="red",
                 )
 
-        # Emit tool usage finished event
-        crewai_event_bus.emit(
-            self,
-            event=ToolUsageFinishedEvent(
-                output=result,
-                tool_name=func_name,
-                tool_args=args_dict,
-                from_agent=self.agent,
-                from_task=self.task,
-                agent_key=agent_key,
-                started_at=started_at,
-                finished_at=datetime.now(),
-            ),
-        )
+        if not error_event_emitted:
+            crewai_event_bus.emit(
+                self,
+                event=ToolUsageFinishedEvent(
+                    output=result,
+                    tool_name=func_name,
+                    tool_args=args_dict,
+                    from_agent=self.agent,
+                    from_task=self.task,
+                    agent_key=agent_key,
+                    started_at=started_at,
+                    finished_at=datetime.now(),
+                ),
+            )
 
         # Append tool result message
         tool_message: LLMMessage = {
@@ -970,18 +987,7 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
         Returns:
             Dictionary with agent output.
         """
-        if "system" in self.prompt:
-            system_prompt = self._format_prompt(
-                cast(str, self.prompt.get("system", "")), inputs
-            )
-            user_prompt = self._format_prompt(
-                cast(str, self.prompt.get("user", "")), inputs
-            )
-            self.messages.append(format_message_for_llm(system_prompt, role="system"))
-            self.messages.append(format_message_for_llm(user_prompt))
-        else:
-            user_prompt = self._format_prompt(self.prompt.get("prompt", ""), inputs)
-            self.messages.append(format_message_for_llm(user_prompt))
+        self._setup_messages(inputs)
 
         await self._ainject_multimodal_files(inputs)
 
@@ -1003,7 +1009,7 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
             raise
 
         if self.ask_for_human_input:
-            formatted_answer = self._handle_human_feedback(formatted_answer)
+            formatted_answer = await self._ahandle_human_feedback(formatted_answer)
 
         self._create_short_term_memory(formatted_answer)
         self._create_long_term_memory(formatted_answer)
@@ -1491,7 +1497,7 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
         return prompt.replace("{tools}", inputs["tools"])
 
     def _handle_human_feedback(self, formatted_answer: AgentFinish) -> AgentFinish:
-        """Process human feedback.
+        """Process human feedback via the configured provider.
 
         Args:
             formatted_answer: Initial agent result.
@@ -1499,17 +1505,22 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
         Returns:
             Final answer after feedback.
         """
-        output_str = (
-            formatted_answer.output
-            if isinstance(formatted_answer.output, str)
-            else formatted_answer.output.model_dump_json()
-        )
-        human_feedback = self._ask_human_input(output_str)
+        provider = get_provider()
+        return provider.handle_feedback(formatted_answer, self)
 
-        if self._is_training_mode():
-            return self._handle_training_feedback(formatted_answer, human_feedback)
+    async def _ahandle_human_feedback(
+        self, formatted_answer: AgentFinish
+    ) -> AgentFinish:
+        """Process human feedback asynchronously via the configured provider.
 
-        return self._handle_regular_feedback(formatted_answer, human_feedback)
+        Args:
+            formatted_answer: Initial agent result.
+
+        Returns:
+            Final answer after feedback.
+        """
+        provider = get_provider()
+        return await provider.handle_feedback_async(formatted_answer, self)
 
     def _is_training_mode(self) -> bool:
         """Check if training mode is active.
@@ -1519,74 +1530,18 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
         """
         return bool(self.crew and self.crew._train)
 
-    def _handle_training_feedback(
-        self, initial_answer: AgentFinish, feedback: str
-    ) -> AgentFinish:
-        """Process training feedback.
+    def _format_feedback_message(self, feedback: str) -> LLMMessage:
+        """Format feedback as a message for the LLM.
 
         Args:
-            initial_answer: Initial agent output.
-            feedback: Training feedback.
+            feedback: User feedback string.
 
         Returns:
-            Improved answer.
+            Formatted message dict.
         """
-        self._handle_crew_training_output(initial_answer, feedback)
-        self.messages.append(
-            format_message_for_llm(
-                self._i18n.slice("feedback_instructions").format(feedback=feedback)
-            )
+        return format_message_for_llm(
+            self._i18n.slice("feedback_instructions").format(feedback=feedback)
         )
-        improved_answer = self._invoke_loop()
-        self._handle_crew_training_output(improved_answer)
-        self.ask_for_human_input = False
-        return improved_answer
-
-    def _handle_regular_feedback(
-        self, current_answer: AgentFinish, initial_feedback: str
-    ) -> AgentFinish:
-        """Process regular feedback iteratively.
-
-        Args:
-            current_answer: Current agent output.
-            initial_feedback: Initial user feedback.
-
-        Returns:
-            Final answer after iterations.
-        """
-        feedback = initial_feedback
-        answer = current_answer
-
-        while self.ask_for_human_input:
-            # If the user provides a blank response, assume they are happy with the result
-            if feedback.strip() == "":
-                self.ask_for_human_input = False
-            else:
-                answer = self._process_feedback_iteration(feedback)
-                output_str = (
-                    answer.output
-                    if isinstance(answer.output, str)
-                    else answer.output.model_dump_json()
-                )
-                feedback = self._ask_human_input(output_str)
-
-        return answer
-
-    def _process_feedback_iteration(self, feedback: str) -> AgentFinish:
-        """Process single feedback iteration.
-
-        Args:
-            feedback: User feedback.
-
-        Returns:
-            Updated agent response.
-        """
-        self.messages.append(
-            format_message_for_llm(
-                self._i18n.slice("feedback_instructions").format(feedback=feedback)
-            )
-        )
-        return self._invoke_loop()
 
     @classmethod
     def __get_pydantic_core_schema__(

lib/crewai/src/crewai/cli/constants.py

@@ -1,10 +1,13 @@
+from typing import Any
+
+
 DEFAULT_CREWAI_ENTERPRISE_URL = "https://app.crewai.com"
 CREWAI_ENTERPRISE_DEFAULT_OAUTH2_PROVIDER = "workos"
 CREWAI_ENTERPRISE_DEFAULT_OAUTH2_AUDIENCE = "client_01JNJQWBJ4SPFN3SWJM5T7BDG8"
 CREWAI_ENTERPRISE_DEFAULT_OAUTH2_CLIENT_ID = "client_01JYT06R59SP0NXYGD994NFXXX"
 CREWAI_ENTERPRISE_DEFAULT_OAUTH2_DOMAIN = "login.crewai.com"
 
-ENV_VARS = {
+ENV_VARS: dict[str, list[dict[str, Any]]] = {
     "openai": [
         {
             "prompt": "Enter your OPENAI API key (press Enter to skip)",
@@ -112,7 +115,7 @@ ENV_VARS = {
 }
 
 
-PROVIDERS = [
+PROVIDERS: list[str] = [
     "openai",
     "anthropic",
     "gemini",
@@ -127,7 +130,7 @@ PROVIDERS = [
     "sambanova",
 ]
 
-MODELS = {
+MODELS: dict[str, list[str]] = {
     "openai": [
         "gpt-4",
         "gpt-4.1",

lib/crewai/src/crewai/cli/create_crew.py

@@ -3,6 +3,7 @@ import shutil
 import sys
 
 import click
+import tomli
 
 from crewai.cli.constants import ENV_VARS, MODELS
 from crewai.cli.provider import (
@@ -13,7 +14,31 @@ from crewai.cli.provider import (
 from crewai.cli.utils import copy_template, load_env_vars, write_env_file
 
 
-def create_folder_structure(name, parent_folder=None):
+def get_reserved_script_names() -> set[str]:
+    """Get reserved script names from pyproject.toml template.
+
+    Returns:
+        Set of reserved script names that would conflict with crew folder names.
+    """
+    package_dir = Path(__file__).parent
+    template_path = package_dir / "templates" / "crew" / "pyproject.toml"
+
+    with open(template_path, "r") as f:
+        template_content = f.read()
+
+    template_content = template_content.replace("{{folder_name}}", "_placeholder_")
+    template_content = template_content.replace("{{name}}", "placeholder")
+    template_content = template_content.replace("{{crew_name}}", "Placeholder")
+
+    template_data = tomli.loads(template_content)
+    script_names = set(template_data.get("project", {}).get("scripts", {}).keys())
+    script_names.discard("_placeholder_")
+    return script_names
+
+
+def create_folder_structure(
+    name: str, parent_folder: str | None = None
+) -> tuple[Path, str, str]:
     import keyword
     import re
 
@@ -51,6 +76,14 @@ def create_folder_structure(name, parent_folder=None):
             f"Project name '{name}' would generate invalid Python module name '{folder_name}'"
         )
 
+    reserved_names = get_reserved_script_names()
+    if folder_name in reserved_names:
+        raise ValueError(
+            f"Project name '{name}' would generate folder name '{folder_name}' which is reserved. "
+            f"Reserved names are: {', '.join(sorted(reserved_names))}. "
+            "Please choose a different name."
+        )
+
     class_name = name.replace("_", " ").replace("-", " ").title().replace(" ", "")
 
     class_name = re.sub(r"[^a-zA-Z0-9_]", "", class_name)
@@ -114,7 +147,9 @@ def create_folder_structure(name, parent_folder=None):
     return folder_path, folder_name, class_name
 
 
-def copy_template_files(folder_path, name, class_name, parent_folder):
+def copy_template_files(
+    folder_path: Path, name: str, class_name: str, parent_folder: str | None
+) -> None:
     package_dir = Path(__file__).parent
     templates_dir = package_dir / "templates" / "crew"
 
@@ -155,7 +190,12 @@ def copy_template_files(folder_path, name, class_name, parent_folder):
             copy_template(src_file, dst_file, name, class_name, folder_path.name)
 
 
-def create_crew(name, provider=None, skip_provider=False, parent_folder=None):
+def create_crew(
+    name: str,
+    provider: str | None = None,
+    skip_provider: bool = False,
+    parent_folder: str | None = None,
+) -> None:
     folder_path, folder_name, class_name = create_folder_structure(name, parent_folder)
     env_vars = load_env_vars(folder_path)
     if not skip_provider:
@@ -189,7 +229,9 @@ def create_crew(name, provider=None, skip_provider=False, parent_folder=None):
             if selected_provider is None:  # User typed 'q'
                 click.secho("Exiting...", fg="yellow")
                 sys.exit(0)
-            if selected_provider:  # Valid selection
+            if selected_provider and isinstance(
+                selected_provider, str
+            ):  # Valid selection
                 break
             click.secho(
                 "No provider selected. Please try again or press 'q' to exit.", fg="red"

lib/crewai/src/crewai/cli/plus_api.py

@@ -1,6 +1,8 @@
+import os
 from typing import Any
 from urllib.parse import urljoin
-import os
+
+import httpx
 import requests
 
 from crewai.cli.config import Settings
@@ -33,7 +35,11 @@ class PlusAPI:
         if settings.org_uuid:
             self.headers["X-Crewai-Organization-Id"] = settings.org_uuid
 
-        self.base_url = os.getenv("CREWAI_PLUS_URL") or str(settings.enterprise_base_url) or DEFAULT_CREWAI_ENTERPRISE_URL
+        self.base_url = (
+            os.getenv("CREWAI_PLUS_URL")
+            or str(settings.enterprise_base_url)
+            or DEFAULT_CREWAI_ENTERPRISE_URL
+        )
 
     def _make_request(
         self, method: str, endpoint: str, **kwargs: Any
@@ -49,8 +55,10 @@ class PlusAPI:
     def get_tool(self, handle: str) -> requests.Response:
         return self._make_request("GET", f"{self.TOOLS_RESOURCE}/{handle}")
 
-    def get_agent(self, handle: str) -> requests.Response:
-        return self._make_request("GET", f"{self.AGENTS_RESOURCE}/{handle}")
+    async def get_agent(self, handle: str) -> httpx.Response:
+        url = urljoin(self.base_url, f"{self.AGENTS_RESOURCE}/{handle}")
+        async with httpx.AsyncClient() as client:
+            return await client.get(url, headers=self.headers)
 
     def publish_tool(
         self,

lib/crewai/src/crewai/cli/provider.py

@@ -1,8 +1,10 @@
 from collections import defaultdict
+from collections.abc import Sequence
 import json
 import os
 from pathlib import Path
 import time
+from typing import Any
 
 import certifi
 import click
@@ -11,16 +13,15 @@ import requests
 from crewai.cli.constants import JSON_URL, MODELS, PROVIDERS
 
 
-def select_choice(prompt_message, choices):
-    """
-    Presents a list of choices to the user and prompts them to select one.
+def select_choice(prompt_message: str, choices: Sequence[str]) -> str | None:
+    """Presents a list of choices to the user and prompts them to select one.
 
     Args:
-    - prompt_message (str): The message to display to the user before presenting the choices.
-    - choices (list): A list of options to present to the user.
+        prompt_message: The message to display to the user before presenting the choices.
+        choices: A list of options to present to the user.
 
     Returns:
-    - str: The selected choice from the list, or None if the user chooses to quit.
+        The selected choice from the list, or None if the user chooses to quit.
     """
 
     provider_models = get_provider_data()
@@ -52,16 +53,14 @@ def select_choice(prompt_message, choices):
         )
 
 
-def select_provider(provider_models):
-    """
-    Presents a list of providers to the user and prompts them to select one.
+def select_provider(provider_models: dict[str, list[str]]) -> str | None | bool:
+    """Presents a list of providers to the user and prompts them to select one.
 
     Args:
-    - provider_models (dict): A dictionary of provider models.
+        provider_models: A dictionary of provider models.
 
     Returns:
-    - str: The selected provider
-    - None: If user explicitly quits
+        The selected provider, None if user explicitly quits, or False if no selection.
     """
     predefined_providers = [p.lower() for p in PROVIDERS]
     all_providers = sorted(set(predefined_providers + list(provider_models.keys())))
@@ -80,16 +79,15 @@ def select_provider(provider_models):
     return provider.lower() if provider else False
 
 
-def select_model(provider, provider_models):
-    """
-    Presents a list of models for a given provider to the user and prompts them to select one.
+def select_model(provider: str, provider_models: dict[str, list[str]]) -> str | None:
+    """Presents a list of models for a given provider to the user and prompts them to select one.
 
     Args:
-    - provider (str): The provider for which to select a model.
-    - provider_models (dict): A dictionary of provider models.
+        provider: The provider for which to select a model.
+        provider_models: A dictionary of provider models.
 
     Returns:
-    - str: The selected model, or None if the operation is aborted or an invalid selection is made.
+        The selected model, or None if the operation is aborted or an invalid selection is made.
     """
     predefined_providers = [p.lower() for p in PROVIDERS]
 
@@ -107,16 +105,17 @@ def select_model(provider, provider_models):
     )
 
 
-def load_provider_data(cache_file, cache_expiry):
-    """
-    Loads provider data from a cache file if it exists and is not expired. If the cache is expired or corrupted, it fetches the data from the web.
+def load_provider_data(cache_file: Path, cache_expiry: int) -> dict[str, Any] | None:
+    """Loads provider data from a cache file if it exists and is not expired.
+
+    If the cache is expired or corrupted, it fetches the data from the web.
 
     Args:
-    - cache_file (Path): The path to the cache file.
-    - cache_expiry (int): The cache expiry time in seconds.
+        cache_file: The path to the cache file.
+        cache_expiry: The cache expiry time in seconds.
 
     Returns:
-    - dict or None: The loaded provider data or None if the operation fails.
+        The loaded provider data or None if the operation fails.
     """
     current_time = time.time()
     if (
@@ -137,32 +136,31 @@ def load_provider_data(cache_file, cache_expiry):
     return fetch_provider_data(cache_file)
 
 
-def read_cache_file(cache_file):
-    """
-    Reads and returns the JSON content from a cache file. Returns None if the file contains invalid JSON.
+def read_cache_file(cache_file: Path) -> dict[str, Any] | None:
+    """Reads and returns the JSON content from a cache file.
 
     Args:
-    - cache_file (Path): The path to the cache file.
+        cache_file: The path to the cache file.
 
     Returns:
-    - dict or None: The JSON content of the cache file or None if the JSON is invalid.
+        The JSON content of the cache file or None if the JSON is invalid.
     """
     try:
         with open(cache_file, "r") as f:
-            return json.load(f)
+            data: dict[str, Any] = json.load(f)
+            return data
     except json.JSONDecodeError:
         return None
 
 
-def fetch_provider_data(cache_file):
-    """
-    Fetches provider data from a specified URL and caches it to a file.
+def fetch_provider_data(cache_file: Path) -> dict[str, Any] | None:
+    """Fetches provider data from a specified URL and caches it to a file.
 
     Args:
-    - cache_file (Path): The path to the cache file.
+        cache_file: The path to the cache file.
 
     Returns:
-    - dict or None: The fetched provider data or None if the operation fails.
+        The fetched provider data or None if the operation fails.
     """
     ssl_config = os.environ["SSL_CERT_FILE"] = certifi.where()
 
@@ -180,36 +178,39 @@ def fetch_provider_data(cache_file):
     return None
 
 
-def download_data(response):
-    """
-    Downloads data from a given HTTP response and returns the JSON content.
+def download_data(response: requests.Response) -> dict[str, Any]:
+    """Downloads data from a given HTTP response and returns the JSON content.
 
     Args:
-    - response (requests.Response): The HTTP response object.
+        response: The HTTP response object.
 
     Returns:
-    - dict: The JSON content of the response.
+        The JSON content of the response.
     """
     total_size = int(response.headers.get("content-length", 0))
     block_size = 8192
-    data_chunks = []
+    data_chunks: list[bytes] = []
+    bar: Any
     with click.progressbar(
         length=total_size, label="Downloading", show_pos=True
-    ) as progress_bar:
+    ) as bar:
         for chunk in response.iter_content(block_size):
             if chunk:
                 data_chunks.append(chunk)
-                progress_bar.update(len(chunk))
+                bar.update(len(chunk))
     data_content = b"".join(data_chunks)
-    return json.loads(data_content.decode("utf-8"))
+    result: dict[str, Any] = json.loads(data_content.decode("utf-8"))
+    return result
 
 
-def get_provider_data():
-    """
-    Retrieves provider data from a cache file, filters out models based on provider criteria, and returns a dictionary of providers mapped to their models.
+def get_provider_data() -> dict[str, list[str]] | None:
+    """Retrieves provider data from a cache file.
+
+    Filters out models based on provider criteria, and returns a dictionary of providers
+    mapped to their models.
 
     Returns:
-    - dict or None: A dictionary of providers mapped to their models or None if the operation fails.
+        A dictionary of providers mapped to their models or None if the operation fails.
     """
     cache_dir = Path.home() / ".crewai"
     cache_dir.mkdir(exist_ok=True)

lib/crewai/src/crewai/cli/version.py

@@ -1,6 +1,107 @@
+"""Version utilities for CrewAI CLI."""
+
+from collections.abc import Mapping
+from datetime import datetime, timedelta
+from functools import lru_cache
 import importlib.metadata
+import json
+from pathlib import Path
+from typing import Any, cast
+from urllib import request
+from urllib.error import URLError
+
+import appdirs
+from packaging.version import InvalidVersion, parse
+
+
+@lru_cache(maxsize=1)
+def _get_cache_file() -> Path:
+    """Get the path to the version cache file.
+
+    Cached to avoid repeated filesystem operations.
+    """
+    cache_dir = Path(appdirs.user_cache_dir("crewai"))
+    cache_dir.mkdir(parents=True, exist_ok=True)
+    return cache_dir / "version_cache.json"
 
 
 def get_crewai_version() -> str:
-    """Get the version number of CrewAI running the CLI"""
+    """Get the version number of CrewAI running the CLI."""
     return importlib.metadata.version("crewai")
+
+
+def _is_cache_valid(cache_data: Mapping[str, Any]) -> bool:
+    """Check if the cache is still valid, less than 24 hours old."""
+    if "timestamp" not in cache_data:
+        return False
+
+    try:
+        cache_time = datetime.fromisoformat(str(cache_data["timestamp"]))
+        return datetime.now() - cache_time < timedelta(hours=24)
+    except (ValueError, TypeError):
+        return False
+
+
+def get_latest_version_from_pypi(timeout: int = 2) -> str | None:
+    """Get the latest version of CrewAI from PyPI.
+
+    Args:
+        timeout: Request timeout in seconds.
+
+    Returns:
+        Latest version string or None if unable to fetch.
+    """
+    cache_file = _get_cache_file()
+    if cache_file.exists():
+        try:
+            cache_data = json.loads(cache_file.read_text())
+            if _is_cache_valid(cache_data):
+                return cast(str | None, cache_data.get("version"))
+        except (json.JSONDecodeError, OSError):
+            pass
+
+    try:
+        with request.urlopen(
+            "https://pypi.org/pypi/crewai/json", timeout=timeout
+        ) as response:
+            data = json.loads(response.read())
+            latest_version = cast(str, data["info"]["version"])
+
+            cache_data = {
+                "version": latest_version,
+                "timestamp": datetime.now().isoformat(),
+            }
+            cache_file.write_text(json.dumps(cache_data))
+
+            return latest_version
+    except (URLError, json.JSONDecodeError, KeyError, OSError):
+        return None
+
+
+def check_version() -> tuple[str, str | None]:
+    """Check current and latest versions.
+
+    Returns:
+        Tuple of (current_version, latest_version).
+        latest_version is None if unable to fetch from PyPI.
+    """
+    current = get_crewai_version()
+    latest = get_latest_version_from_pypi()
+    return current, latest
+
+
+def is_newer_version_available() -> tuple[bool, str, str | None]:
+    """Check if a newer version is available.
+
+    Returns:
+        Tuple of (is_newer, current_version, latest_version).
+    """
+    current, latest = check_version()
+
+    if latest is None:
+        return False, current, None
+
+    try:
+        return parse(latest) > parse(current), current, latest
+    except (InvalidVersion, TypeError):
+        return False, current, latest

lib/crewai/src/crewai/context.py

@@ -43,3 +43,23 @@ def platform_context(integration_token: str) -> Generator[None, Any, None]:
         yield
     finally:
         _platform_integration_token.reset(token)
+
+
+_current_task_id: contextvars.ContextVar[str | None] = contextvars.ContextVar(
+    "current_task_id", default=None
+)
+
+
+def set_current_task_id(task_id: str | None) -> contextvars.Token[str | None]:
+    """Set the current task ID in the context. Returns a token for reset."""
+    return _current_task_id.set(task_id)
+
+
+def reset_current_task_id(token: contextvars.Token[str | None]) -> None:
+    """Reset the current task ID to its previous value."""
+    _current_task_id.reset(token)
+
+
+def get_current_task_id() -> str | None:
+    """Get the current task ID from the context."""
+    return _current_task_id.get()

lib/crewai/src/crewai/core/__init__.py

@@ -0,0 +1 @@
+"""Core crewAI components and interfaces."""

lib/crewai/src/crewai/core/providers/__init__.py

@@ -0,0 +1 @@
+"""Provider interfaces for extensible crewAI components."""

lib/crewai/src/crewai/core/providers/content_processor.py

@@ -0,0 +1,78 @@
+"""Content processor provider for extensible content processing."""
+
+from contextvars import ContextVar
+from typing import Any, Protocol, runtime_checkable
+
+
+@runtime_checkable
+class ContentProcessorProvider(Protocol):
+    """Protocol for content processing during task execution."""
+
+    def process(self, content: str, context: dict[str, Any] | None = None) -> str:
+        """Process content before use.
+
+        Args:
+            content: The content to process.
+            context: Optional context information.
+
+        Returns:
+            The processed content.
+        """
+        ...
+
+
+class NoOpContentProcessor:
+    """Default processor that returns content unchanged."""
+
+    def process(self, content: str, context: dict[str, Any] | None = None) -> str:
+        """Return content unchanged.
+
+        Args:
+            content: The content to process.
+            context: Optional context information (unused).
+
+        Returns:
+            The original content unchanged.
+        """
+        return content
+
+
+_content_processor: ContextVar[ContentProcessorProvider | None] = ContextVar(
+    "_content_processor", default=None
+)
+
+_default_processor = NoOpContentProcessor()
+
+
+def get_processor() -> ContentProcessorProvider:
+    """Get the current content processor.
+
+    Returns:
+        The registered content processor or the default no-op processor.
+    """
+    processor = _content_processor.get()
+    if processor is not None:
+        return processor
+    return _default_processor
+
+
+def set_processor(processor: ContentProcessorProvider) -> None:
+    """Set the content processor for the current context.
+
+    Args:
+        processor: The content processor to use.
+    """
+    _content_processor.set(processor)
+
+
+def process_content(content: str, context: dict[str, Any] | None = None) -> str:
+    """Process content using the registered processor.
+
+    Args:
+        content: The content to process.
+        context: Optional context information.
+
+    Returns:
+        The processed content.
+    """
+    return get_processor().process(content, context)

lib/crewai/src/crewai/core/providers/human_input.py

@@ -0,0 +1,489 @@
+"""Human input provider for HITL (Human-in-the-Loop) flows."""
+
+from __future__ import annotations
+
+import asyncio
+from contextvars import ContextVar, Token
+import sys
+from typing import TYPE_CHECKING, Protocol, runtime_checkable
+
+
+if TYPE_CHECKING:
+    from crewai.agent.core import Agent
+    from crewai.agents.parser import AgentFinish
+    from crewai.crew import Crew
+    from crewai.llms.base_llm import BaseLLM
+    from crewai.task import Task
+    from crewai.utilities.types import LLMMessage
+
+
+class ExecutorContext(Protocol):
+    """Context interface for human input providers to interact with executor."""
+
+    task: Task | None
+    crew: Crew | None
+    messages: list[LLMMessage]
+    ask_for_human_input: bool
+    llm: BaseLLM
+    agent: Agent
+
+    def _invoke_loop(self) -> AgentFinish:
+        """Invoke the agent loop and return the result."""
+        ...
+
+    def _is_training_mode(self) -> bool:
+        """Check if training mode is active."""
+        ...
+
+    def _handle_crew_training_output(
+        self,
+        result: AgentFinish,
+        human_feedback: str | None = None,
+    ) -> None:
+        """Handle training output."""
+        ...
+
+    def _format_feedback_message(self, feedback: str) -> LLMMessage:
+        """Format feedback as a message."""
+        ...
+
+
+class AsyncExecutorContext(ExecutorContext, Protocol):
+    """Extended context for executors that support async invocation."""
+
+    async def _ainvoke_loop(self) -> AgentFinish:
+        """Invoke the agent loop asynchronously and return the result."""
+        ...
+
+
+@runtime_checkable
+class HumanInputProvider(Protocol):
+    """Protocol for human input handling.
+
+    Implementations handle the full feedback flow:
+    - Sync: prompt user, loop until satisfied
+    - Async: use non-blocking I/O and async invoke loop
+    """
+
+    def setup_messages(self, context: ExecutorContext) -> bool:
+        """Set up messages for execution.
+
+        Called before standard message setup. Allows providers to handle
+        conversation resumption or other custom message initialization.
+
+        Args:
+            context: Executor context with messages list to modify.
+
+        Returns:
+            True if messages were set up (skip standard setup),
+            False to use standard setup.
+        """
+        ...
+
+    def post_setup_messages(self, context: ExecutorContext) -> None:
+        """Called after standard message setup.
+
+        Allows providers to modify messages after standard setup completes.
+        Only called when setup_messages returned False.
+
+        Args:
+            context: Executor context with messages list to modify.
+        """
+        ...
+
+    def handle_feedback(
+        self,
+        formatted_answer: AgentFinish,
+        context: ExecutorContext,
+    ) -> AgentFinish:
+        """Handle the full human feedback flow synchronously.
+
+        Args:
+            formatted_answer: The agent's current answer.
+            context: Executor context for callbacks.
+
+        Returns:
+            The final answer after feedback processing.
+
+        Raises:
+            Exception: Async implementations may raise to signal external handling.
+        """
+        ...
+
+    async def handle_feedback_async(
+        self,
+        formatted_answer: AgentFinish,
+        context: AsyncExecutorContext,
+    ) -> AgentFinish:
+        """Handle the full human feedback flow asynchronously.
+
+        Uses non-blocking I/O for user prompts and async invoke loop
+        for agent re-execution.
+
+        Args:
+            formatted_answer: The agent's current answer.
+            context: Async executor context for callbacks.
+
+        Returns:
+            The final answer after feedback processing.
+        """
+        ...
+
+    @staticmethod
+    def _get_output_string(answer: AgentFinish) -> str:
+        """Extract output string from answer.
+
+        Args:
+            answer: The agent's finished answer.
+
+        Returns:
+            String representation of the output.
+        """
+        if isinstance(answer.output, str):
+            return answer.output
+        return answer.output.model_dump_json()
+
+
+class SyncHumanInputProvider(HumanInputProvider):
+    """Default human input provider with sync and async support."""
+
+    def setup_messages(self, context: ExecutorContext) -> bool:
+        """Use standard message setup.
+
+        Args:
+            context: Executor context (unused).
+
+        Returns:
+            False to use standard setup.
+        """
+        return False
+
+    def post_setup_messages(self, context: ExecutorContext) -> None:
+        """No-op for sync provider.
+
+        Args:
+            context: Executor context (unused).
+        """
+
+    def handle_feedback(
+        self,
+        formatted_answer: AgentFinish,
+        context: ExecutorContext,
+    ) -> AgentFinish:
+        """Handle feedback synchronously with terminal prompts.
+
+        Args:
+            formatted_answer: The agent's current answer.
+            context: Executor context for callbacks.
+
+        Returns:
+            The final answer after feedback processing.
+        """
+        feedback = self._prompt_input(context.crew)
+
+        if context._is_training_mode():
+            return self._handle_training_feedback(formatted_answer, feedback, context)
+
+        return self._handle_regular_feedback(formatted_answer, feedback, context)
+
+    async def handle_feedback_async(
+        self,
+        formatted_answer: AgentFinish,
+        context: AsyncExecutorContext,
+    ) -> AgentFinish:
+        """Handle feedback asynchronously without blocking the event loop.
+
+        Args:
+            formatted_answer: The agent's current answer.
+            context: Async executor context for callbacks.
+
+        Returns:
+            The final answer after feedback processing.
+        """
+        feedback = await self._prompt_input_async(context.crew)
+
+        if context._is_training_mode():
+            return await self._handle_training_feedback_async(
+                formatted_answer, feedback, context
+            )
+
+        return await self._handle_regular_feedback_async(
+            formatted_answer, feedback, context
+        )
+
+    # ── Sync helpers ──────────────────────────────────────────────────
+
+    @staticmethod
+    def _handle_training_feedback(
+        initial_answer: AgentFinish,
+        feedback: str,
+        context: ExecutorContext,
+    ) -> AgentFinish:
+        """Process training feedback (single iteration).
+
+        Args:
+            initial_answer: The agent's initial answer.
+            feedback: Human feedback string.
+            context: Executor context for callbacks.
+
+        Returns:
+            Improved answer after processing feedback.
+        """
+        context._handle_crew_training_output(initial_answer, feedback)
+        context.messages.append(context._format_feedback_message(feedback))
+        improved_answer = context._invoke_loop()
+        context._handle_crew_training_output(improved_answer)
+        context.ask_for_human_input = False
+        return improved_answer
+
+    def _handle_regular_feedback(
+        self,
+        current_answer: AgentFinish,
+        initial_feedback: str,
+        context: ExecutorContext,
+    ) -> AgentFinish:
+        """Process regular feedback with iteration loop.
+
+        Args:
+            current_answer: The agent's current answer.
+            initial_feedback: Initial human feedback string.
+            context: Executor context for callbacks.
+
+        Returns:
+            Final answer after all feedback iterations.
+        """
+        feedback = initial_feedback
+        answer = current_answer
+
+        while context.ask_for_human_input:
+            if feedback.strip() == "":
+                context.ask_for_human_input = False
+            else:
+                context.messages.append(context._format_feedback_message(feedback))
+                answer = context._invoke_loop()
+                feedback = self._prompt_input(context.crew)
+
+        return answer
+
+    # ── Async helpers ─────────────────────────────────────────────────
+
+    @staticmethod
+    async def _handle_training_feedback_async(
+        initial_answer: AgentFinish,
+        feedback: str,
+        context: AsyncExecutorContext,
+    ) -> AgentFinish:
+        """Process training feedback asynchronously (single iteration).
+
+        Args:
+            initial_answer: The agent's initial answer.
+            feedback: Human feedback string.
+            context: Async executor context for callbacks.
+
+        Returns:
+            Improved answer after processing feedback.
+        """
+        context._handle_crew_training_output(initial_answer, feedback)
+        context.messages.append(context._format_feedback_message(feedback))
+        improved_answer = await context._ainvoke_loop()
+        context._handle_crew_training_output(improved_answer)
+        context.ask_for_human_input = False
+        return improved_answer
+
+    async def _handle_regular_feedback_async(
+        self,
+        current_answer: AgentFinish,
+        initial_feedback: str,
+        context: AsyncExecutorContext,
+    ) -> AgentFinish:
+        """Process regular feedback with async iteration loop.
+
+        Args:
+            current_answer: The agent's current answer.
+            initial_feedback: Initial human feedback string.
+            context: Async executor context for callbacks.
+
+        Returns:
+            Final answer after all feedback iterations.
+        """
+        feedback = initial_feedback
+        answer = current_answer
+
+        while context.ask_for_human_input:
+            if feedback.strip() == "":
+                context.ask_for_human_input = False
+            else:
+                context.messages.append(context._format_feedback_message(feedback))
+                answer = await context._ainvoke_loop()
+                feedback = await self._prompt_input_async(context.crew)
+
+        return answer
+
+    # ── I/O ───────────────────────────────────────────────────────────
+
+    @staticmethod
+    def _prompt_input(crew: Crew | None) -> str:
+        """Show rich panel and prompt for input.
+
+        Args:
+            crew: The crew instance for context.
+
+        Returns:
+            User input string from terminal.
+        """
+        from rich.panel import Panel
+        from rich.text import Text
+
+        from crewai.events.event_listener import event_listener
+
+        formatter = event_listener.formatter
+        formatter.pause_live_updates()
+
+        try:
+            if crew and getattr(crew, "_train", False):
+                prompt_text = (
+                    "TRAINING MODE: Provide feedback to improve the agent's performance.\n\n"
+                    "This will be used to train better versions of the agent.\n"
+                    "Please provide detailed feedback about the result quality and reasoning process."
+                )
+                title = "🎓 Training Feedback Required"
+            else:
+                prompt_text = (
+                    "Provide feedback on the Final Result above.\n\n"
+                    "• If you are happy with the result, simply hit Enter without typing anything.\n"
+                    "• Otherwise, provide specific improvement requests.\n"
+                    "• You can provide multiple rounds of feedback until satisfied."
+                )
+                title = "💬 Human Feedback Required"
+
+            content = Text()
+            content.append(prompt_text, style="yellow")
+
+            prompt_panel = Panel(
+                content,
+                title=title,
+                border_style="yellow",
+                padding=(1, 2),
+            )
+            formatter.console.print(prompt_panel)
+
+            response = input()
+            if response.strip() != "":
+                formatter.console.print("\n[cyan]Processing your feedback...[/cyan]")
+            return response
+        finally:
+            formatter.resume_live_updates()
+
+    @staticmethod
+    async def _prompt_input_async(crew: Crew | None) -> str:
+        """Show rich panel and prompt for input without blocking the event loop.
+
+        Args:
+            crew: The crew instance for context.
+
+        Returns:
+            User input string from terminal.
+        """
+        from rich.panel import Panel
+        from rich.text import Text
+
+        from crewai.events.event_listener import event_listener
+
+        formatter = event_listener.formatter
+        formatter.pause_live_updates()
+
+        try:
+            if crew and getattr(crew, "_train", False):
+                prompt_text = (
+                    "TRAINING MODE: Provide feedback to improve the agent's performance.\n\n"
+                    "This will be used to train better versions of the agent.\n"
+                    "Please provide detailed feedback about the result quality and reasoning process."
+                )
+                title = "🎓 Training Feedback Required"
+            else:
+                prompt_text = (
+                    "Provide feedback on the Final Result above.\n\n"
+                    "• If you are happy with the result, simply hit Enter without typing anything.\n"
+                    "• Otherwise, provide specific improvement requests.\n"
+                    "• You can provide multiple rounds of feedback until satisfied."
+                )
+                title = "💬 Human Feedback Required"
+
+            content = Text()
+            content.append(prompt_text, style="yellow")
+
+            prompt_panel = Panel(
+                content,
+                title=title,
+                border_style="yellow",
+                padding=(1, 2),
+            )
+            formatter.console.print(prompt_panel)
+
+            response = await _async_readline()
+            if response.strip() != "":
+                formatter.console.print("\n[cyan]Processing your feedback...[/cyan]")
+            return response
+        finally:
+            formatter.resume_live_updates()
+
+
+async def _async_readline() -> str:
+    """Read a line from stdin using the event loop's native I/O.
+
+    Falls back to asyncio.to_thread on platforms where piping stdin
+    is unsupported.
+
+    Returns:
+        The line read from stdin, with trailing newline stripped.
+    """
+    loop = asyncio.get_running_loop()
+    try:
+        reader = asyncio.StreamReader()
+        protocol = asyncio.StreamReaderProtocol(reader)
+        await loop.connect_read_pipe(lambda: protocol, sys.stdin)
+        raw = await reader.readline()
+        return raw.decode().rstrip("\n")
+    except (OSError, NotImplementedError, ValueError):
+        return await asyncio.to_thread(input)
+
+
+_provider: ContextVar[HumanInputProvider | None] = ContextVar(
+    "human_input_provider",
+    default=None,
+)
+
+
+def get_provider() -> HumanInputProvider:
+    """Get the current human input provider.
+
+    Returns:
+        The current provider, or a new SyncHumanInputProvider if none set.
+    """
+    provider = _provider.get()
+    if provider is None:
+        initialized_provider = SyncHumanInputProvider()
+        set_provider(initialized_provider)
+        return initialized_provider
+    return provider
+
+
+def set_provider(provider: HumanInputProvider) -> Token[HumanInputProvider | None]:
+    """Set the human input provider for the current context.
+
+    Args:
+        provider: The provider to use.
+
+    Returns:
+        Token that can be used to reset to previous value.
+    """
+    return _provider.set(provider)
+
+
+def reset_provider(token: Token[HumanInputProvider | None]) -> None:
+    """Reset the provider to its previous value.
+
+    Args:
+        token: Token returned from set_provider.
+    """
+    _provider.reset(token)

lib/crewai/src/crewai/crew.py

@@ -751,6 +751,8 @@ class Crew(FlowTrackable, BaseModel):
             for after_callback in self.after_kickoff_callbacks:
                 result = after_callback(result)
 
+            result = self._post_kickoff(result)
+
             self.usage_metrics = self.calculate_usage_metrics()
 
             return result
@@ -764,6 +766,9 @@ class Crew(FlowTrackable, BaseModel):
             clear_files(self.id)
             detach(token)
 
+    def _post_kickoff(self, result: CrewOutput) -> CrewOutput:
+        return result
+
     def kickoff_for_each(
         self,
         inputs: list[dict[str, Any]],
@@ -936,6 +941,8 @@ class Crew(FlowTrackable, BaseModel):
             for after_callback in self.after_kickoff_callbacks:
                 result = after_callback(result)
 
+            result = self._post_kickoff(result)
+
             self.usage_metrics = self.calculate_usage_metrics()
 
             return result
@@ -1181,6 +1188,9 @@ class Crew(FlowTrackable, BaseModel):
             self.manager_agent = manager
         manager.crew = self
 
+    def _get_execution_start_index(self, tasks: list[Task]) -> int | None:
+        return None
+
     def _execute_tasks(
         self,
         tasks: list[Task],
@@ -1197,6 +1207,9 @@ class Crew(FlowTrackable, BaseModel):
         Returns:
             CrewOutput: Final output of the crew
         """
+        custom_start = self._get_execution_start_index(tasks)
+        if custom_start is not None:
+            start_index = custom_start
 
         task_outputs: list[TaskOutput] = []
         futures: list[tuple[Task, Future[TaskOutput], int]] = []
@@ -1305,8 +1318,10 @@ class Crew(FlowTrackable, BaseModel):
         if files:
             supported_types: list[str] = []
             if agent and agent.llm and agent.llm.supports_multimodal():
-                provider = getattr(agent.llm, "provider", None) or getattr(
-                    agent.llm, "model", "openai"
+                provider = (
+                    getattr(agent.llm, "provider", None)
+                    or getattr(agent.llm, "model", None)
+                    or "openai"
                 )
                 api = getattr(agent.llm, "api", None)
                 supported_types = get_supported_content_types(provider, api)
@@ -1502,6 +1517,7 @@ class Crew(FlowTrackable, BaseModel):
         final_string_output = final_task_output.raw
         self._finish_execution(final_string_output)
         self.token_usage = self.calculate_usage_metrics()
+        crewai_event_bus.flush()
         crewai_event_bus.emit(
             self,
             CrewKickoffCompletedEvent(
@@ -2011,7 +2027,13 @@ class Crew(FlowTrackable, BaseModel):
     @staticmethod
     def _show_tracing_disabled_message() -> None:
         """Show a message when tracing is disabled."""
-        from crewai.events.listeners.tracing.utils import has_user_declined_tracing
+        from crewai.events.listeners.tracing.utils import (
+            has_user_declined_tracing,
+            should_suppress_tracing_messages,
+        )
+
+        if should_suppress_tracing_messages():
+            return
 
         console = Console()
 

lib/crewai/src/crewai/events/__init__.py

@@ -195,6 +195,7 @@ __all__ = [
     "ToolUsageFinishedEvent",
     "ToolUsageStartedEvent",
     "ToolValidateInputErrorEvent",
+    "_extension_exports",
     "crewai_event_bus",
 ]
 
@@ -210,14 +211,29 @@ _AGENT_EVENT_MAPPING = {
     "LiteAgentExecutionStartedEvent": "crewai.events.types.agent_events",
 }
 
+_extension_exports: dict[str, Any] = {}
+
 
 def __getattr__(name: str) -> Any:
-    """Lazy import for agent events to avoid circular imports."""
+    """Lazy import for agent events and registered extensions."""
     if name in _AGENT_EVENT_MAPPING:
         import importlib
 
         module_path = _AGENT_EVENT_MAPPING[name]
         module = importlib.import_module(module_path)
         return getattr(module, name)
+
+    if name in _extension_exports:
+        import importlib
+
+        value = _extension_exports[name]
+        if isinstance(value, str):
+            module_path, _, attr_name = value.rpartition(".")
+            if module_path:
+                module = importlib.import_module(module_path)
+                return getattr(module, attr_name)
+            return importlib.import_module(value)
+        return value
+
     msg = f"module {__name__!r} has no attribute {name!r}"
     raise AttributeError(msg)

lib/crewai/src/crewai/events/base_events.py

@@ -63,6 +63,7 @@ class BaseEvent(BaseModel):
     parent_event_id: str | None = None
     previous_event_id: str | None = None
     triggered_by_event_id: str | None = None
+    started_event_id: str | None = None
     emission_sequence: int | None = None
 
     def to_json(self, exclude: set[str] | None = None) -> Serializable:

lib/crewai/src/crewai/events/event_bus.py

@@ -227,6 +227,39 @@ class CrewAIEventsBus:
 
         return decorator
 
+    def off(
+        self,
+        event_type: type[BaseEvent],
+        handler: Callable[..., Any],
+    ) -> None:
+        """Unregister an event handler for a specific event type.
+
+        Args:
+            event_type: The event class to stop listening for
+            handler: The handler function to unregister
+        """
+        with self._rwlock.w_locked():
+            if event_type in self._sync_handlers:
+                existing_sync = self._sync_handlers[event_type]
+                if handler in existing_sync:
+                    self._sync_handlers[event_type] = existing_sync - {handler}
+                    if not self._sync_handlers[event_type]:
+                        del self._sync_handlers[event_type]
+
+            if event_type in self._async_handlers:
+                existing_async = self._async_handlers[event_type]
+                if handler in existing_async:
+                    self._async_handlers[event_type] = existing_async - {handler}
+                    if not self._async_handlers[event_type]:
+                        del self._async_handlers[event_type]
+
+            if event_type in self._handler_dependencies:
+                self._handler_dependencies[event_type].pop(handler, None)
+                if not self._handler_dependencies[event_type]:
+                    del self._handler_dependencies[event_type]
+
+            self._execution_plan_cache.pop(event_type, None)
+
     def _call_handlers(
         self,
         source: Any,
@@ -374,7 +407,8 @@ class CrewAIEventsBus:
                 if popped is None:
                     handle_empty_pop(event_type_name)
                 else:
-                    _, popped_type = popped
+                    popped_event_id, popped_type = popped
+                    event.started_event_id = popped_event_id
                     expected_start = VALID_EVENT_PAIRS.get(event_type_name)
                     if expected_start and popped_type and popped_type != expected_start:
                         handle_mismatch(event_type_name, popped_type, expected_start)
@@ -536,24 +570,52 @@ class CrewAIEventsBus:
             ...     # Do stuff...
             ... # Handlers are cleared after the context
         """
-        with self._rwlock.w_locked():
-            prev_sync = self._sync_handlers
-            prev_async = self._async_handlers
-            prev_deps = self._handler_dependencies
-            prev_cache = self._execution_plan_cache
-            self._sync_handlers = {}
-            self._async_handlers = {}
-            self._handler_dependencies = {}
-            self._execution_plan_cache = {}
+        with self._rwlock.r_locked():
+            saved_sync: dict[type[BaseEvent], frozenset[SyncHandler]] = dict(
+                self._sync_handlers
+            )
+            saved_async: dict[type[BaseEvent], frozenset[AsyncHandler]] = dict(
+                self._async_handlers
+            )
+            saved_deps: dict[type[BaseEvent], dict[Handler, list[Depends[Any]]]] = {
+                event_type: dict(handlers)
+                for event_type, handlers in self._handler_dependencies.items()
+            }
+
+        for event_type, sync_handlers in saved_sync.items():
+            for sync_handler in sync_handlers:
+                self.off(event_type, sync_handler)
+
+        for event_type, async_handlers in saved_async.items():
+            for async_handler in async_handlers:
+                self.off(event_type, async_handler)
 
         try:
             yield
         finally:
-            with self._rwlock.w_locked():
-                self._sync_handlers = prev_sync
-                self._async_handlers = prev_async
-                self._handler_dependencies = prev_deps
-                self._execution_plan_cache = prev_cache
+            with self._rwlock.r_locked():
+                current_sync = dict(self._sync_handlers)
+                current_async = dict(self._async_handlers)
+
+            for event_type, cur_sync in current_sync.items():
+                orig_sync = saved_sync.get(event_type, frozenset())
+                for new_handler in cur_sync - orig_sync:
+                    self.off(event_type, new_handler)
+
+            for event_type, cur_async in current_async.items():
+                orig_async = saved_async.get(event_type, frozenset())
+                for new_async_handler in cur_async - orig_async:
+                    self.off(event_type, new_async_handler)
+
+            for event_type, sync_handlers in saved_sync.items():
+                for sync_handler in sync_handlers:
+                    deps = saved_deps.get(event_type, {}).get(sync_handler)
+                    self._register_handler(event_type, sync_handler, deps)
+
+            for event_type, async_handlers in saved_async.items():
+                for async_handler in async_handlers:
+                    deps = saved_deps.get(event_type, {}).get(async_handler)
+                    self._register_handler(event_type, async_handler, deps)
 
     def shutdown(self, wait: bool = True) -> None:
         """Gracefully shutdown the event loop and wait for all tasks to finish.

lib/crewai/src/crewai/events/listeners/tracing/trace_listener.py

@@ -797,7 +797,13 @@ class TraceCollectionListener(BaseEventListener):
         from rich.console import Console
         from rich.panel import Panel
 
-        from crewai.events.listeners.tracing.utils import has_user_declined_tracing
+        from crewai.events.listeners.tracing.utils import (
+            has_user_declined_tracing,
+            should_suppress_tracing_messages,
+        )
+
+        if should_suppress_tracing_messages():
+            return
 
         console = Console()
 

lib/crewai/src/crewai/events/listeners/tracing/utils.py

@@ -1,3 +1,4 @@
+from collections.abc import Callable
 from contextvars import ContextVar, Token
 from datetime import datetime
 import getpass
@@ -26,6 +27,35 @@ logger = logging.getLogger(__name__)
 
 _tracing_enabled: ContextVar[bool | None] = ContextVar("_tracing_enabled", default=None)
 
+_first_time_trace_hook: ContextVar[Callable[[], bool] | None] = ContextVar(
+    "_first_time_trace_hook", default=None
+)
+
+_suppress_tracing_messages: ContextVar[bool] = ContextVar(
+    "_suppress_tracing_messages", default=False
+)
+
+
+def set_suppress_tracing_messages(suppress: bool) -> object:
+    """Set whether to suppress tracing-related console messages.
+
+    Args:
+        suppress: True to suppress messages, False to show them.
+
+    Returns:
+        A token that can be used to restore the previous value.
+    """
+    return _suppress_tracing_messages.set(suppress)
+
+
+def should_suppress_tracing_messages() -> bool:
+    """Check if tracing messages should be suppressed.
+
+    Returns:
+        True if messages should be suppressed, False otherwise.
+    """
+    return _suppress_tracing_messages.get()
+
 
 def should_enable_tracing(*, override: bool | None = None) -> bool:
     """Determine if tracing should be enabled.
@@ -407,10 +437,13 @@ def truncate_messages(
 def should_auto_collect_first_time_traces() -> bool:
     """True if we should auto-collect traces for first-time user.
 
-
     Returns:
         True if first-time user AND telemetry not disabled AND tracing not explicitly enabled, False otherwise.
     """
+    hook = _first_time_trace_hook.get()
+    if hook is not None:
+        return hook()
+
     if _is_test_environment():
         return False
 
@@ -432,6 +465,9 @@ def prompt_user_for_trace_viewing(timeout_seconds: int = 20) -> bool:
     if _is_test_environment():
         return False
 
+    if should_suppress_tracing_messages():
+        return False
+
     try:
         import threading
 

lib/crewai/src/crewai/events/types/llm_events.py

@@ -10,6 +10,7 @@ class LLMEventBase(BaseEvent):
     from_task: Any | None = None
     from_agent: Any | None = None
     model: str | None = None
+    call_id: str
 
     def __init__(self, **data: Any) -> None:
         if data.get("from_task"):

lib/crewai/src/crewai/events/types/tool_usage_events.py

@@ -16,7 +16,7 @@ class ToolUsageEvent(BaseEvent):
     tool_name: str
     tool_args: dict[str, Any] | str
     tool_class: str | None = None
-    run_attempts: int | None = None
+    run_attempts: int = 0
     delegations: int | None = None
     agent: Any | None = None
     task_name: str | None = None
@@ -26,7 +26,7 @@ class ToolUsageEvent(BaseEvent):
 
     model_config = ConfigDict(arbitrary_types_allowed=True)
 
-    def __init__(self, **data):
+    def __init__(self, **data: Any) -> None:
         if data.get("from_task"):
             task = data["from_task"]
             data["task_id"] = str(task.id)
@@ -96,10 +96,10 @@ class ToolExecutionErrorEvent(BaseEvent):
     type: str = "tool_execution_error"
     tool_name: str
     tool_args: dict[str, Any]
-    tool_class: Callable
+    tool_class: Callable[..., Any]
     agent: Any | None = None
 
-    def __init__(self, **data):
+    def __init__(self, **data: Any) -> None:
         super().__init__(**data)
         # Set fingerprint data from the agent
         if self.agent and hasattr(self.agent, "fingerprint") and self.agent.fingerprint:

lib/crewai/src/crewai/events/utils/console_formatter.py

@@ -1,11 +1,45 @@
+from contextvars import ContextVar
+import os
 import threading
-from typing import Any, ClassVar
+from typing import Any, ClassVar, cast
 
 from rich.console import Console
 from rich.live import Live
 from rich.panel import Panel
 from rich.text import Text
 
+from crewai.cli.version import is_newer_version_available
+
+
+_disable_version_check: ContextVar[bool] = ContextVar(
+    "_disable_version_check", default=False
+)
+
+_suppress_console_output: ContextVar[bool] = ContextVar(
+    "_suppress_console_output", default=False
+)
+
+
+def set_suppress_console_output(suppress: bool) -> object:
+    """Set whether to suppress all console output.
+
+    Args:
+        suppress: True to suppress output, False to show it.
+
+    Returns:
+        A token that can be used to restore the previous value.
+    """
+    return _suppress_console_output.set(suppress)
+
+
+def should_suppress_console_output() -> bool:
+    """Check if console output should be suppressed.
+
+    Returns:
+        True if output should be suppressed, False otherwise.
+    """
+    return _suppress_console_output.get()
+
 
 class ConsoleFormatter:
     tool_usage_counts: ClassVar[dict[str, int]] = {}
@@ -35,13 +69,56 @@ class ConsoleFormatter:
             padding=(1, 2),
         )
 
+    def _show_version_update_message_if_needed(self) -> None:
+        """Show version update message if a newer version is available.
+
+        Only displays when verbose mode is enabled and not running in CI/CD.
+        """
+        if not self.verbose:
+            return
+
+        if _disable_version_check.get():
+            return
+
+        if os.getenv("CI", "").lower() in ("true", "1"):
+            return
+
+        if os.getenv("CREWAI_DISABLE_VERSION_CHECK", "").lower() in ("true", "1"):
+            return
+
+        try:
+            is_newer, current, latest = is_newer_version_available()
+            if is_newer and latest:
+                message = f"""A new version of CrewAI is available!
+
+Current version: {current}
+Latest version:  {latest}
+
+To update, run: uv sync --upgrade-package crewai"""
+
+                panel = Panel(
+                    message,
+                    title="✨ Update Available ✨",
+                    border_style="yellow",
+                    padding=(1, 2),
+                )
+                self.console.print(panel)
+                self.console.print()
+        except Exception:  # noqa: S110
+            # Silently ignore errors in version check - it's non-critical
+            pass
+
     def _show_tracing_disabled_message_if_needed(self) -> None:
         """Show tracing disabled message if tracing is not enabled."""
         from crewai.events.listeners.tracing.utils import (
             has_user_declined_tracing,
             is_tracing_enabled_in_context,
+            should_suppress_tracing_messages,
         )
 
+        if should_suppress_tracing_messages():
+            return
+
         if not is_tracing_enabled_in_context():
             if has_user_declined_tracing():
                 message = """Info: Tracing is disabled.
@@ -93,6 +170,8 @@ To enable tracing, do any one of these:
 
     def print(self, *args: Any, **kwargs: Any) -> None:
         """Print to console. Simplified to only handle panel-based output."""
+        if should_suppress_console_output():
+            return
         # Skip blank lines during streaming
         if len(args) == 0 and self._is_streaming:
             return
@@ -176,9 +255,10 @@ To enable tracing, do any one of these:
         if not self.verbose:
             return
 
-        # Reset the crew completion event for this new crew execution
         ConsoleFormatter.crew_completion_printed.clear()
 
+        self._show_version_update_message_if_needed()
+
         content = self.create_status_content(
             "Crew Execution Started",
             crew_name,
@@ -237,6 +317,8 @@ To enable tracing, do any one of these:
 
     def handle_flow_started(self, flow_name: str, flow_id: str) -> None:
         """Show flow started panel."""
+        self._show_version_update_message_if_needed()
+
         content = Text()
         content.append("Flow Started\n", style="blue bold")
         content.append("Name: ", style="white")
@@ -446,6 +528,9 @@ To enable tracing, do any one of these:
         if not self.verbose:
             return
 
+        if should_suppress_console_output():
+            return
+
         self._is_streaming = True
         self._last_stream_call_type = call_type
 
@@ -885,7 +970,7 @@ To enable tracing, do any one of these:
 
             is_a2a_delegation = False
             try:
-                output_data = json.loads(formatted_answer.output)
+                output_data = json.loads(cast(str, formatted_answer.output))
                 if isinstance(output_data, dict):
                     if output_data.get("is_a2a") is True:
                         is_a2a_delegation = True

lib/crewai/src/crewai/experimental/agent_executor.py

@@ -18,6 +18,7 @@ from crewai.agents.parser import (
     AgentFinish,
     OutputParserError,
 )
+from crewai.core.providers.human_input import get_provider
 from crewai.events.event_bus import crewai_event_bus
 from crewai.events.listeners.tracing.utils import (
     is_tracing_enabled_in_context,
@@ -31,7 +32,8 @@ from crewai.events.types.tool_usage_events import (
     ToolUsageFinishedEvent,
     ToolUsageStartedEvent,
 )
-from crewai.flow.flow import Flow, listen, or_, router, start
+from crewai.flow.flow import Flow, StateProxy, listen, or_, router, start
+from crewai.flow.types import FlowMethodName
 from crewai.hooks.llm_hooks import (
     get_after_llm_call_hooks,
     get_before_llm_call_hooks,
@@ -41,7 +43,12 @@ from crewai.hooks.tool_hooks import (
     get_after_tool_call_hooks,
     get_before_tool_call_hooks,
 )
-from crewai.hooks.types import AfterLLMCallHookType, BeforeLLMCallHookType
+from crewai.hooks.types import (
+    AfterLLMCallHookCallable,
+    AfterLLMCallHookType,
+    BeforeLLMCallHookCallable,
+    BeforeLLMCallHookType,
+)
 from crewai.utilities.agent_utils import (
     convert_tools_to_openai_schema,
     enforce_rpm_limit,
@@ -191,8 +198,12 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
 
         self._instance_id = str(uuid4())[:8]
 
-        self.before_llm_call_hooks: list[BeforeLLMCallHookType] = []
-        self.after_llm_call_hooks: list[AfterLLMCallHookType] = []
+        self.before_llm_call_hooks: list[
+            BeforeLLMCallHookType | BeforeLLMCallHookCallable
+        ] = []
+        self.after_llm_call_hooks: list[
+            AfterLLMCallHookType | AfterLLMCallHookCallable
+        ] = []
         self.before_llm_call_hooks.extend(get_before_llm_call_hooks())
         self.after_llm_call_hooks.extend(get_after_llm_call_hooks())
 
@@ -207,6 +218,71 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
             )
         self._state = AgentReActState()
 
+    @property
+    def messages(self) -> list[LLMMessage]:
+        """Delegate to state for ExecutorContext conformance."""
+        return self._state.messages
+
+    @messages.setter
+    def messages(self, value: list[LLMMessage]) -> None:
+        """Delegate to state for ExecutorContext conformance."""
+        if self._flow_initialized and hasattr(self, "_state_lock"):
+            with self._state_lock:
+                self._state.messages = value
+        else:
+            self._state.messages = value
+
+    @property
+    def ask_for_human_input(self) -> bool:
+        """Delegate to state for ExecutorContext conformance."""
+        return self._state.ask_for_human_input
+
+    @ask_for_human_input.setter
+    def ask_for_human_input(self, value: bool) -> None:
+        """Delegate to state for ExecutorContext conformance."""
+        self._state.ask_for_human_input = value
+
+    def _invoke_loop(self) -> AgentFinish:
+        """Invoke the agent loop and return the result.
+
+        Required by ExecutorContext protocol.
+        """
+        self._state.iterations = 0
+        self._state.is_finished = False
+        self._state.current_answer = None
+
+        self.kickoff()
+
+        answer = self._state.current_answer
+        if not isinstance(answer, AgentFinish):
+            raise RuntimeError("Agent loop did not produce a final answer")
+        return answer
+
+    async def _ainvoke_loop(self) -> AgentFinish:
+        """Invoke the agent loop asynchronously and return the result.
+
+        Required by AsyncExecutorContext protocol.
+        """
+        self._state.iterations = 0
+        self._state.is_finished = False
+        self._state.current_answer = None
+
+        await self.akickoff()
+
+        answer = self._state.current_answer
+        if not isinstance(answer, AgentFinish):
+            raise RuntimeError("Agent loop did not produce a final answer")
+        return answer
+
+    def _format_feedback_message(self, feedback: str) -> LLMMessage:
+        """Format feedback as a message for the LLM.
+
+        Required by ExecutorContext protocol.
+        """
+        return format_message_for_llm(
+            self._i18n.slice("feedback_instructions").format(feedback=feedback)
+        )
+
     def _ensure_flow_initialized(self) -> None:
         """Ensure Flow.__init__() has been called.
 
@@ -298,18 +374,10 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
         Flow initialization is deferred to prevent event emission during agent setup.
         Returns the temporary state until invoke() is called.
         """
+        if self._flow_initialized and hasattr(self, "_state_lock"):
+            return StateProxy(self._state, self._state_lock)  # type: ignore[return-value]
         return self._state
 
-    @property
-    def messages(self) -> list[LLMMessage]:
-        """Compatibility property for mixin - returns state messages."""
-        return self._state.messages
-
-    @messages.setter
-    def messages(self, value: list[LLMMessage]) -> None:
-        """Set state messages."""
-        self._state.messages = value
-
     @property
     def iterations(self) -> int:
         """Compatibility property for mixin - returns state iterations."""
@@ -416,15 +484,14 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
             raise
 
     @listen("continue_reasoning_native")
-    def call_llm_native_tools(
-        self,
-    ) -> Literal["native_tool_calls", "native_finished", "context_error"]:
+    def call_llm_native_tools(self) -> None:
         """Execute LLM call with native function calling.
 
         Always calls the LLM so it can read reflection prompts and decide
         whether to provide a final answer or request more tools.
 
-        Returns routing decision based on whether tool calls or final answer.
+        Note: This is a listener, not a router. The route_native_tool_result
+        router fires after this to determine the next step based on state.
         """
         try:
             # Clear pending tools - LLM will decide what to do next after reading
@@ -454,8 +521,7 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
             if isinstance(answer, list) and answer and self._is_tool_call_list(answer):
                 # Store tool calls for sequential processing
                 self.state.pending_tool_calls = list(answer)
-
-                return "native_tool_calls"
+                return  # Router will check pending_tool_calls
 
             if isinstance(answer, BaseModel):
                 self.state.current_answer = AgentFinish(
@@ -465,7 +531,7 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
                 )
                 self._invoke_step_callback(self.state.current_answer)
                 self._append_message_to_state(answer.model_dump_json())
-                return "native_finished"
+                return  # Router will check current_answer
 
             # Text response - this is the final answer
             if isinstance(answer, str):
@@ -476,8 +542,7 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
                 )
                 self._invoke_step_callback(self.state.current_answer)
                 self._append_message_to_state(answer)
-
-                return "native_finished"
+                return  # Router will check current_answer
 
             # Unexpected response type, treat as final answer
             self.state.current_answer = AgentFinish(
@@ -487,13 +552,12 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
             )
             self._invoke_step_callback(self.state.current_answer)
             self._append_message_to_state(str(answer))
-
-            return "native_finished"
+            # Router will check current_answer
 
         except Exception as e:
             if is_context_length_exceeded(e):
                 self._last_context_error = e
-                return "context_error"
+                return  # Router will check _last_context_error
             if e.__class__.__module__.startswith("litellm"):
                 raise e
             handle_unknown_error(self._printer, e, verbose=self.agent.verbose)
@@ -506,6 +570,22 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
             return "execute_tool"
         return "agent_finished"
 
+    @router(call_llm_native_tools)
+    def route_native_tool_result(
+        self,
+    ) -> Literal["native_tool_calls", "native_finished", "context_error"]:
+        """Route based on LLM response for native tool calling.
+
+        Checks state set by call_llm_native_tools to determine next step.
+        This router is needed because only router return values trigger
+        downstream listeners.
+        """
+        if self._last_context_error is not None:
+            return "context_error"
+        if self.state.pending_tool_calls:
+            return "native_tool_calls"
+        return "native_finished"
+
     @listen("execute_tool")
     def execute_tool_action(self) -> Literal["tool_completed", "tool_result_is_final"]:
         """Execute the tool action and handle the result."""
@@ -689,6 +769,7 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
                     agent_key=agent_key,
                 ),
             )
+            error_event_emitted = False
 
             track_delegation_if_needed(func_name, args_dict, self.task)
 
@@ -764,6 +845,7 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
                                 error=e,
                             ),
                         )
+                        error_event_emitted = True
             elif max_usage_reached and original_tool:
                 # Return error message when max usage limit is reached
                 result = f"Tool '{func_name}' has reached its usage limit of {original_tool.max_usage_count} times and cannot be used anymore."
@@ -792,20 +874,20 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
                         color="red",
                     )
 
-            # Emit tool usage finished event
-            crewai_event_bus.emit(
-                self,
-                event=ToolUsageFinishedEvent(
-                    output=result,
-                    tool_name=func_name,
-                    tool_args=args_dict,
-                    from_agent=self.agent,
-                    from_task=self.task,
-                    agent_key=agent_key,
-                    started_at=started_at,
-                    finished_at=datetime.now(),
-                ),
-            )
+            if not error_event_emitted:
+                crewai_event_bus.emit(
+                    self,
+                    event=ToolUsageFinishedEvent(
+                        output=result,
+                        tool_name=func_name,
+                        tool_args=args_dict,
+                        from_agent=self.agent,
+                        from_task=self.task,
+                        agent_key=agent_key,
+                        started_at=started_at,
+                        finished_at=datetime.now(),
+                    ),
+                )
 
             # Append tool result message
             tool_message: LLMMessage = {
@@ -861,9 +943,11 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
         self.state.iterations += 1
         return "initialized"
 
-    @listen("initialized")
+    @listen(or_("initialized", "tool_completed", "native_tool_completed"))
     def continue_iteration(self) -> Literal["check_iteration"]:
         """Bridge listener that connects iteration loop back to iteration check."""
+        if self._flow_initialized:
+            self._discard_or_listener(FlowMethodName("continue_iteration"))
         return "check_iteration"
 
     @router(or_(initialize_reasoning, continue_iteration))
@@ -1105,7 +1189,7 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
                 )
 
             if self.state.ask_for_human_input:
-                formatted_answer = self._handle_human_feedback(formatted_answer)
+                formatted_answer = await self._ahandle_human_feedback(formatted_answer)
 
             self._create_short_term_memory(formatted_answer)
             self._create_long_term_memory(formatted_answer)
@@ -1319,17 +1403,22 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
         Returns:
             Final answer after feedback.
         """
-        output_str = (
-            str(formatted_answer.output)
-            if isinstance(formatted_answer.output, BaseModel)
-            else formatted_answer.output
-        )
-        human_feedback = self._ask_human_input(output_str)
+        provider = get_provider()
+        return provider.handle_feedback(formatted_answer, self)
 
-        if self._is_training_mode():
-            return self._handle_training_feedback(formatted_answer, human_feedback)
+    async def _ahandle_human_feedback(
+        self, formatted_answer: AgentFinish
+    ) -> AgentFinish:
+        """Process human feedback asynchronously and refine answer.
 
-        return self._handle_regular_feedback(formatted_answer, human_feedback)
+        Args:
+            formatted_answer: Initial agent result.
+
+        Returns:
+            Final answer after feedback.
+        """
+        provider = get_provider()
+        return await provider.handle_feedback_async(formatted_answer, self)
 
     def _is_training_mode(self) -> bool:
         """Check if training mode is active.
@@ -1339,101 +1428,6 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
         """
         return bool(self.crew and self.crew._train)
 
-    def _handle_training_feedback(
-        self, initial_answer: AgentFinish, feedback: str
-    ) -> AgentFinish:
-        """Process training feedback and generate improved answer.
-
-        Args:
-            initial_answer: Initial agent output.
-            feedback: Training feedback.
-
-        Returns:
-            Improved answer.
-        """
-        self._handle_crew_training_output(initial_answer, feedback)
-        self.state.messages.append(
-            format_message_for_llm(
-                self._i18n.slice("feedback_instructions").format(feedback=feedback)
-            )
-        )
-
-        # Re-run flow for improved answer
-        self.state.iterations = 0
-        self.state.is_finished = False
-        self.state.current_answer = None
-
-        self.kickoff()
-
-        # Get improved answer from state
-        improved_answer = self.state.current_answer
-        if not isinstance(improved_answer, AgentFinish):
-            raise RuntimeError(
-                "Training feedback iteration did not produce final answer"
-            )
-
-        self._handle_crew_training_output(improved_answer)
-        self.state.ask_for_human_input = False
-        return improved_answer
-
-    def _handle_regular_feedback(
-        self, current_answer: AgentFinish, initial_feedback: str
-    ) -> AgentFinish:
-        """Process regular feedback iteratively until user is satisfied.
-
-        Args:
-            current_answer: Current agent output.
-            initial_feedback: Initial user feedback.
-
-        Returns:
-            Final answer after iterations.
-        """
-        feedback = initial_feedback
-        answer = current_answer
-
-        while self.state.ask_for_human_input:
-            if feedback.strip() == "":
-                self.state.ask_for_human_input = False
-            else:
-                answer = self._process_feedback_iteration(feedback)
-                output_str = (
-                    str(answer.output)
-                    if isinstance(answer.output, BaseModel)
-                    else answer.output
-                )
-                feedback = self._ask_human_input(output_str)
-
-        return answer
-
-    def _process_feedback_iteration(self, feedback: str) -> AgentFinish:
-        """Process a single feedback iteration and generate updated response.
-
-        Args:
-            feedback: User feedback.
-
-        Returns:
-            Updated agent response.
-        """
-        self.state.messages.append(
-            format_message_for_llm(
-                self._i18n.slice("feedback_instructions").format(feedback=feedback)
-            )
-        )
-
-        # Re-run flow
-        self.state.iterations = 0
-        self.state.is_finished = False
-        self.state.current_answer = None
-
-        self.kickoff()
-
-        # Get answer from state
-        answer = self.state.current_answer
-        if not isinstance(answer, AgentFinish):
-            raise RuntimeError("Feedback iteration did not produce final answer")
-
-        return answer
-
     @classmethod
     def __get_pydantic_core_schema__(
         cls, _source_type: Any, _handler: GetCoreSchemaHandler

lib/crewai/src/crewai/flow/async_feedback/__init__.py

@@ -8,11 +8,13 @@ Example:
     from crewai.flow import Flow, start, human_feedback
     from crewai.flow.async_feedback import HumanFeedbackProvider, HumanFeedbackPending
 
+
     class SlackProvider(HumanFeedbackProvider):
         def request_feedback(self, context, flow):
             self.send_slack_notification(context)
             raise HumanFeedbackPending(context=context)
 
+
     class MyFlow(Flow):
         @start()
         @human_feedback(
@@ -26,16 +28,30 @@ Example:
     ```
 """
 
+from typing import Any
+
+from crewai.flow.async_feedback.providers import ConsoleProvider
 from crewai.flow.async_feedback.types import (
     HumanFeedbackPending,
     HumanFeedbackProvider,
     PendingFeedbackContext,
 )
-from crewai.flow.async_feedback.providers import ConsoleProvider
+
 
 __all__ = [
     "ConsoleProvider",
     "HumanFeedbackPending",
     "HumanFeedbackProvider",
     "PendingFeedbackContext",
+    "_extension_exports",
 ]
+
+_extension_exports: dict[str, Any] = {}
+
+
+def __getattr__(name: str) -> Any:
+    """Support extensions via dynamic attribute lookup."""
+    if name in _extension_exports:
+        return _extension_exports[name]
+    msg = f"module {__name__!r} has no attribute {name!r}"
+    raise AttributeError(msg)

lib/crewai/src/crewai/flow/async_feedback/providers.py

@@ -6,10 +6,11 @@ provider that collects feedback via console input.
 
 from __future__ import annotations
 
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, Any
 
 from crewai.flow.async_feedback.types import PendingFeedbackContext
 
+
 if TYPE_CHECKING:
     from crewai.flow.flow import Flow
 
@@ -27,6 +28,7 @@ class ConsoleProvider:
         ```python
         from crewai.flow.async_feedback import ConsoleProvider
 
+
         # Explicitly use console provider
         @human_feedback(
             message="Review this:",
@@ -49,7 +51,7 @@ class ConsoleProvider:
     def request_feedback(
         self,
         context: PendingFeedbackContext,
-        flow: Flow,
+        flow: Flow[Any],
     ) -> str:
         """Request feedback via console input (blocking).
 

lib/crewai/src/crewai/flow/async_feedback/types.py

@@ -10,6 +10,7 @@ from dataclasses import dataclass, field
 from datetime import datetime
 from typing import TYPE_CHECKING, Any, Protocol, runtime_checkable
 
+
 if TYPE_CHECKING:
     from crewai.flow.flow import Flow
 
@@ -155,7 +156,7 @@ class HumanFeedbackPending(Exception):  # noqa: N818 - Not an error, a control f
                     callback_info={
                         "slack_channel": "#reviews",
                         "thread_id": ticket_id,
-                    }
+                    },
                 )
         ```
     """
@@ -232,7 +233,7 @@ class HumanFeedbackProvider(Protocol):
                     callback_info={
                         "channel": self.channel,
                         "thread_id": thread_id,
-                    }
+                    },
                 )
         ```
     """
@@ -240,7 +241,7 @@ class HumanFeedbackProvider(Protocol):
     def request_feedback(
         self,
         context: PendingFeedbackContext,
-        flow: Flow,
+        flow: Flow[Any],
     ) -> str:
         """Request feedback from a human.
 

lib/crewai/src/crewai/flow/constants.py

@@ -1,4 +1,5 @@
 from typing import Final, Literal
 
+
 AND_CONDITION: Final[Literal["AND"]] = "AND"
 OR_CONDITION: Final[Literal["OR"]] = "OR"

lib/crewai/src/crewai/flow/flow.py

@@ -7,7 +7,14 @@ for building event-driven workflows with conditional execution and routing.
 from __future__ import annotations
 
 import asyncio
-from collections.abc import Callable, Sequence
+from collections.abc import (
+    Callable,
+    ItemsView,
+    Iterator,
+    KeysView,
+    Sequence,
+    ValuesView,
+)
 from concurrent.futures import Future
 import copy
 import inspect
@@ -45,6 +52,7 @@ from crewai.events.listeners.tracing.utils import (
     has_user_declined_tracing,
     set_tracing_enabled,
     should_enable_tracing,
+    should_suppress_tracing_messages,
 )
 from crewai.events.types.flow_events import (
     FlowCreatedEvent,
@@ -58,6 +66,7 @@ from crewai.events.types.flow_events import (
     MethodExecutionStartedEvent,
 )
 from crewai.flow.constants import AND_CONDITION, OR_CONDITION
+from crewai.flow.flow_context import current_flow_id, current_flow_request_id
 from crewai.flow.flow_wrappers import (
     FlowCondition,
     FlowConditions,
@@ -407,6 +416,132 @@ def and_(*conditions: str | FlowCondition | Callable[..., Any]) -> FlowCondition
     return {"type": AND_CONDITION, "conditions": processed_conditions}
 
 
+class LockedListProxy(Generic[T]):
+    """Thread-safe proxy for list operations.
+
+    Wraps a list and uses a lock for all mutating operations.
+    """
+
+    def __init__(self, lst: list[T], lock: threading.Lock) -> None:
+        self._list = lst
+        self._lock = lock
+
+    def append(self, item: T) -> None:
+        with self._lock:
+            self._list.append(item)
+
+    def extend(self, items: list[T]) -> None:
+        with self._lock:
+            self._list.extend(items)
+
+    def insert(self, index: int, item: T) -> None:
+        with self._lock:
+            self._list.insert(index, item)
+
+    def remove(self, item: T) -> None:
+        with self._lock:
+            self._list.remove(item)
+
+    def pop(self, index: int = -1) -> T:
+        with self._lock:
+            return self._list.pop(index)
+
+    def clear(self) -> None:
+        with self._lock:
+            self._list.clear()
+
+    def __setitem__(self, index: int, value: T) -> None:
+        with self._lock:
+            self._list[index] = value
+
+    def __delitem__(self, index: int) -> None:
+        with self._lock:
+            del self._list[index]
+
+    def __getitem__(self, index: int) -> T:
+        return self._list[index]
+
+    def __len__(self) -> int:
+        return len(self._list)
+
+    def __iter__(self) -> Iterator[T]:
+        return iter(self._list)
+
+    def __contains__(self, item: object) -> bool:
+        return item in self._list
+
+    def __repr__(self) -> str:
+        return repr(self._list)
+
+    def __bool__(self) -> bool:
+        return bool(self._list)
+
+
+class LockedDictProxy(Generic[T]):
+    """Thread-safe proxy for dict operations.
+
+    Wraps a dict and uses a lock for all mutating operations.
+    """
+
+    def __init__(self, d: dict[str, T], lock: threading.Lock) -> None:
+        self._dict = d
+        self._lock = lock
+
+    def __setitem__(self, key: str, value: T) -> None:
+        with self._lock:
+            self._dict[key] = value
+
+    def __delitem__(self, key: str) -> None:
+        with self._lock:
+            del self._dict[key]
+
+    def pop(self, key: str, *default: T) -> T:
+        with self._lock:
+            return self._dict.pop(key, *default)
+
+    def update(self, other: dict[str, T]) -> None:
+        with self._lock:
+            self._dict.update(other)
+
+    def clear(self) -> None:
+        with self._lock:
+            self._dict.clear()
+
+    def setdefault(self, key: str, default: T) -> T:
+        with self._lock:
+            return self._dict.setdefault(key, default)
+
+    def __getitem__(self, key: str) -> T:
+        return self._dict[key]
+
+    def __len__(self) -> int:
+        return len(self._dict)
+
+    def __iter__(self) -> Iterator[str]:
+        return iter(self._dict)
+
+    def __contains__(self, key: object) -> bool:
+        return key in self._dict
+
+    def keys(self) -> KeysView[str]:
+        return self._dict.keys()
+
+    def values(self) -> ValuesView[T]:
+        return self._dict.values()
+
+    def items(self) -> ItemsView[str, T]:
+        return self._dict.items()
+
+    def get(self, key: str, default: T | None = None) -> T | None:
+        return self._dict.get(key, default)
+
+    def __repr__(self) -> str:
+        return repr(self._dict)
+
+    def __bool__(self) -> bool:
+        return bool(self._dict)
+
+
 class StateProxy(Generic[T]):
     """Proxy that provides thread-safe access to flow state.
 
@@ -421,7 +556,13 @@ class StateProxy(Generic[T]):
         object.__setattr__(self, "_proxy_lock", lock)
 
     def __getattr__(self, name: str) -> Any:
-        return getattr(object.__getattribute__(self, "_proxy_state"), name)
+        value = getattr(object.__getattribute__(self, "_proxy_state"), name)
+        lock = object.__getattribute__(self, "_proxy_lock")
+        if isinstance(value, list):
+            return LockedListProxy(value, lock)
+        if isinstance(value, dict):
+            return LockedDictProxy(value, lock)
+        return value
 
     def __setattr__(self, name: str, value: Any) -> None:
         if name in ("_proxy_state", "_proxy_lock"):
@@ -1540,6 +1681,13 @@ class Flow(Generic[T], metaclass=FlowMeta):
         ctx = baggage.set_baggage("flow_input_files", input_files or {}, context=ctx)
         flow_token = attach(ctx)
 
+        flow_id_token = None
+        request_id_token = None
+        if current_flow_id.get() is None:
+            flow_id_token = current_flow_id.set(self.flow_id)
+        if current_flow_request_id.get() is None:
+            request_id_token = current_flow_request_id.set(self.flow_id)
+
         try:
             # Reset flow state for fresh execution unless restoring from persistence
             is_restoring = inputs and "id" in inputs and self._persistence is not None
@@ -1584,7 +1732,6 @@ class Flow(Generic[T], metaclass=FlowMeta):
                 reset_emission_counter()
                 reset_last_event_id()
 
-            # Emit FlowStartedEvent and log the start of the flow.
             if not self.suppress_flow_events:
                 future = crewai_event_bus.emit(
                     self,
@@ -1595,7 +1742,10 @@ class Flow(Generic[T], metaclass=FlowMeta):
                     ),
                 )
                 if future:
-                    self._event_futures.append(future)
+                    try:
+                        await asyncio.wrap_future(future)
+                    except Exception:
+                        logger.warning("FlowStartedEvent handler failed", exc_info=True)
                 self._log_flow_event(
                     f"Flow started with ID: {self.flow_id}", color="bold magenta"
                 )
@@ -1687,6 +1837,12 @@ class Flow(Generic[T], metaclass=FlowMeta):
 
             final_output = self._method_outputs[-1] if self._method_outputs else None
 
+            if self._event_futures:
+                await asyncio.gather(
+                    *[asyncio.wrap_future(f) for f in self._event_futures]
+                )
+                self._event_futures.clear()
+
             if not self.suppress_flow_events:
                 future = crewai_event_bus.emit(
                     self,
@@ -1698,13 +1854,12 @@ class Flow(Generic[T], metaclass=FlowMeta):
                     ),
                 )
                 if future:
-                    self._event_futures.append(future)
-
-            if self._event_futures:
-                await asyncio.gather(
-                    *[asyncio.wrap_future(f) for f in self._event_futures]
-                )
-                self._event_futures.clear()
+                    try:
+                        await asyncio.wrap_future(future)
+                    except Exception:
+                        logger.warning(
+                            "FlowFinishedEvent handler failed", exc_info=True
+                        )
 
             if not self.suppress_flow_events:
                 trace_listener = TraceCollectionListener()
@@ -1717,6 +1872,10 @@ class Flow(Generic[T], metaclass=FlowMeta):
 
             return final_output
         finally:
+            if request_id_token is not None:
+                current_flow_request_id.reset(request_id_token)
+            if flow_id_token is not None:
+                current_flow_id.reset(flow_id_token)
             detach(flow_token)
 
     async def akickoff(
@@ -1775,40 +1934,14 @@ class Flow(Generic[T], metaclass=FlowMeta):
             await self._execute_listeners(start_method_name, result, finished_event_id)
             # Then execute listeners for the router result (e.g., "approved")
             router_result_trigger = FlowMethodName(str(result))
-            listeners_for_result = self._find_triggered_methods(
-                router_result_trigger, router_only=False
+            listener_result = (
+                self.last_human_feedback
+                if self.last_human_feedback is not None
+                else result
+            )
+            await self._execute_listeners(
+                router_result_trigger, listener_result, finished_event_id
             )
-            if listeners_for_result:
-                # Pass the HumanFeedbackResult if available
-                listener_result = (
-                    self.last_human_feedback
-                    if self.last_human_feedback is not None
-                    else result
-                )
-                racing_group = self._get_racing_group_for_listeners(
-                    listeners_for_result
-                )
-                if racing_group:
-                    racing_members, _ = racing_group
-                    other_listeners = [
-                        name
-                        for name in listeners_for_result
-                        if name not in racing_members
-                    ]
-                    await self._execute_racing_listeners(
-                        racing_members,
-                        other_listeners,
-                        listener_result,
-                        finished_event_id,
-                    )
-                else:
-                    tasks = [
-                        self._execute_single_listener(
-                            listener_name, listener_result, finished_event_id
-                        )
-                        for listener_name in listeners_for_result
-                    ]
-                    await asyncio.gather(*tasks)
         else:
             await self._execute_listeners(start_method_name, result, finished_event_id)
 
@@ -2014,15 +2147,14 @@ class Flow(Generic[T], metaclass=FlowMeta):
                 router_input = router_result_to_feedback.get(
                     str(current_trigger), current_result
                 )
-                current_triggering_event_id = await self._execute_single_listener(
+                (
+                    router_result,
+                    current_triggering_event_id,
+                ) = await self._execute_single_listener(
                     router_name, router_input, current_triggering_event_id
                 )
-                # After executing router, the router's result is the path
-                router_result = (
-                    self._method_outputs[-1] if self._method_outputs else None
-                )
                 if router_result:  # Only add non-None results
-                    router_results.append(router_result)
+                    router_results.append(FlowMethodName(str(router_result)))
                     # If this was a human_feedback router, map the outcome to the feedback
                     if self.last_human_feedback is not None:
                         router_result_to_feedback[str(router_result)] = (
@@ -2062,12 +2194,14 @@ class Flow(Generic[T], metaclass=FlowMeta):
                             racing_members,
                             other_listeners,
                             listener_result,
-                            triggering_event_id,
+                            current_triggering_event_id,
                         )
                     else:
                         tasks = [
                             self._execute_single_listener(
-                                listener_name, listener_result, triggering_event_id
+                                listener_name,
+                                listener_result,
+                                current_triggering_event_id,
                             )
                             for listener_name in listeners_triggered
                         ]
@@ -2250,7 +2384,7 @@ class Flow(Generic[T], metaclass=FlowMeta):
         listener_name: FlowMethodName,
         result: Any,
         triggering_event_id: str | None = None,
-    ) -> str | None:
+    ) -> tuple[Any, str | None]:
         """Executes a single listener method with proper event handling.
 
         This internal method manages the execution of an individual listener,
@@ -2263,8 +2397,9 @@ class Flow(Generic[T], metaclass=FlowMeta):
                 used for causal chain tracking.
 
         Returns:
-            The event_id of the MethodExecutionFinishedEvent emitted by this listener,
-            or None if events are suppressed.
+            A tuple of (listener_result, event_id) where listener_result is the return
+            value of the listener method and event_id is the MethodExecutionFinishedEvent
+            id, or (None, None) if skipped during resumption.
 
         Note:
             - Inspects method signature to determine if it accepts the trigger result
@@ -2290,7 +2425,7 @@ class Flow(Generic[T], metaclass=FlowMeta):
                         ):
                             # This conditional start was executed, continue its chain
                             await self._execute_start_method(start_method_name)
-                return None
+                return (None, None)
             # For cyclic flows, clear from completed to allow re-execution
             self._completed_methods.discard(listener_name)
             # Also clear from fired OR listeners for cyclic flows
@@ -2328,46 +2463,7 @@ class Flow(Generic[T], metaclass=FlowMeta):
                 listener_name, listener_result, finished_event_id
             )
 
-            # If this listener is also a router (e.g., has @human_feedback with emit),
-            # we need to trigger listeners for the router result as well
-            if listener_name in self._routers and listener_result is not None:
-                router_result_trigger = FlowMethodName(str(listener_result))
-                listeners_for_result = self._find_triggered_methods(
-                    router_result_trigger, router_only=False
-                )
-                if listeners_for_result:
-                    # Pass the HumanFeedbackResult if available
-                    feedback_result = (
-                        self.last_human_feedback
-                        if self.last_human_feedback is not None
-                        else listener_result
-                    )
-                    racing_group = self._get_racing_group_for_listeners(
-                        listeners_for_result
-                    )
-                    if racing_group:
-                        racing_members, _ = racing_group
-                        other_listeners = [
-                            name
-                            for name in listeners_for_result
-                            if name not in racing_members
-                        ]
-                        await self._execute_racing_listeners(
-                            racing_members,
-                            other_listeners,
-                            feedback_result,
-                            finished_event_id,
-                        )
-                    else:
-                        tasks = [
-                            self._execute_single_listener(
-                                name, feedback_result, finished_event_id
-                            )
-                            for name in listeners_for_result
-                        ]
-                        await asyncio.gather(*tasks)
-
-            return finished_event_id
+            return (listener_result, finished_event_id)
 
         except Exception as e:
             # Don't log HumanFeedbackPending as an error - it's expected control flow
@@ -2614,6 +2710,8 @@ class Flow(Generic[T], metaclass=FlowMeta):
     @staticmethod
     def _show_tracing_disabled_message() -> None:
         """Show a message when tracing is disabled."""
+        if should_suppress_tracing_messages():
+            return
 
         console = Console()
 

lib/crewai/src/crewai/flow/flow_config.py

@@ -8,6 +8,7 @@ from __future__ import annotations
 
 from typing import TYPE_CHECKING, Any
 
+
 if TYPE_CHECKING:
     from crewai.flow.async_feedback.types import HumanFeedbackProvider
 

lib/crewai/src/crewai/flow/flow_context.py

@@ -0,0 +1,16 @@
+"""Flow execution context management.
+
+This module provides context variables for tracking flow execution state across
+async boundaries and nested function calls.
+"""
+
+import contextvars
+
+
+current_flow_request_id: contextvars.ContextVar[str | None] = contextvars.ContextVar(
+    "flow_request_id", default=None
+)
+
+current_flow_id: contextvars.ContextVar[str | None] = contextvars.ContextVar(
+    "flow_id", default=None
+)

lib/crewai/src/crewai/flow/flow_trackable.py

@@ -1,46 +1,22 @@
-import inspect
-from typing import Any
-
-from pydantic import BaseModel, Field, InstanceOf, model_validator
+from pydantic import BaseModel, model_validator
 from typing_extensions import Self
 
-from crewai.flow.flow import Flow
+from crewai.flow.flow_context import current_flow_id, current_flow_request_id
 
 
 class FlowTrackable(BaseModel):
-    """Mixin that tracks the Flow instance that instantiated the object, e.g. a
-    Flow instance that created a Crew or Agent.
+    """Mixin that tracks flow execution context for objects created within flows.
 
-    Automatically finds and stores a reference to the parent Flow instance by
-    inspecting the call stack.
+    When a Crew or Agent is instantiated inside a flow execution, this mixin
+    automatically captures the flow ID and request ID from context variables,
+    enabling proper tracking and association with the parent flow execution.
     """
 
-    parent_flow: InstanceOf[Flow[Any]] | None = Field(
-        default=None,
-        description="The parent flow of the instance, if it was created inside a flow.",
-    )
-
     @model_validator(mode="after")
-    def _set_parent_flow(self) -> Self:
-        max_depth = 8
-        frame = inspect.currentframe()
-
-        try:
-            if frame is None:
-                return self
-
-            frame = frame.f_back
-            for _ in range(max_depth):
-                if frame is None:
-                    break
-
-                candidate = frame.f_locals.get("self")
-                if isinstance(candidate, Flow):
-                    self.parent_flow = candidate
-                    break
-
-                frame = frame.f_back
-        finally:
-            del frame
+    def _set_flow_context(self) -> Self:
+        request_id = current_flow_request_id.get()
+        if request_id:
+            self._request_id = request_id
+            self._flow_id = current_flow_id.get()
 
         return self

lib/crewai/src/crewai/flow/human_feedback.py

@@ -11,6 +11,7 @@ Example (synchronous, default):
     ```python
     from crewai.flow import Flow, start, listen, human_feedback
 
+
     class ReviewFlow(Flow):
         @start()
         @human_feedback(
@@ -32,11 +33,13 @@ Example (asynchronous with custom provider):
     from crewai.flow import Flow, start, human_feedback
     from crewai.flow.async_feedback import HumanFeedbackProvider, HumanFeedbackPending
 
+
     class SlackProvider(HumanFeedbackProvider):
         def request_feedback(self, context, flow):
             self.send_notification(context)
             raise HumanFeedbackPending(context=context)
 
+
     class ReviewFlow(Flow):
         @start()
         @human_feedback(
@@ -229,6 +232,7 @@ def human_feedback(
         def review_document(self):
             return document_content
 
+
         @listen("approved")
         def publish(self):
             print(f"Publishing: {self.last_human_feedback.output}")
@@ -265,7 +269,7 @@ def human_feedback(
     def decorator(func: F) -> F:
         """Inner decorator that wraps the function."""
 
-        def _request_feedback(flow_instance: Flow, method_output: Any) -> str:
+        def _request_feedback(flow_instance: Flow[Any], method_output: Any) -> str:
             """Request feedback using provider or default console."""
             from crewai.flow.async_feedback.types import PendingFeedbackContext
 
@@ -291,19 +295,16 @@ def human_feedback(
                 effective_provider = flow_config.hitl_provider
 
             if effective_provider is not None:
-                # Use provider (may raise HumanFeedbackPending for async providers)
                 return effective_provider.request_feedback(context, flow_instance)
-            else:
-                # Use default console input (local development)
-                return flow_instance._request_human_feedback(
-                    message=message,
-                    output=method_output,
-                    metadata=metadata,
-                    emit=emit,
-                )
+            return flow_instance._request_human_feedback(
+                message=message,
+                output=method_output,
+                metadata=metadata,
+                emit=emit,
+            )
 
         def _process_feedback(
-            flow_instance: Flow,
+            flow_instance: Flow[Any],
             method_output: Any,
             raw_feedback: str,
         ) -> HumanFeedbackResult | str:
@@ -319,12 +320,14 @@ def human_feedback(
                     # No default and no feedback - use first outcome
                     collapsed_outcome = emit[0]
             elif emit:
-                # Collapse feedback to outcome using LLM
-                collapsed_outcome = flow_instance._collapse_to_outcome(
-                    feedback=raw_feedback,
-                    outcomes=emit,
-                    llm=llm,
-                )
+                if llm is not None:
+                    collapsed_outcome = flow_instance._collapse_to_outcome(
+                        feedback=raw_feedback,
+                        outcomes=emit,
+                        llm=llm,
+                    )
+                else:
+                    collapsed_outcome = emit[0]
 
             # Create result
             result = HumanFeedbackResult(
@@ -349,7 +352,7 @@ def human_feedback(
         if asyncio.iscoroutinefunction(func):
             # Async wrapper
             @wraps(func)
-            async def async_wrapper(self: Flow, *args: Any, **kwargs: Any) -> Any:
+            async def async_wrapper(self: Flow[Any], *args: Any, **kwargs: Any) -> Any:
                 # Execute the original method
                 method_output = await func(self, *args, **kwargs)
 
@@ -363,7 +366,7 @@ def human_feedback(
         else:
             # Sync wrapper
             @wraps(func)
-            def sync_wrapper(self: Flow, *args: Any, **kwargs: Any) -> Any:
+            def sync_wrapper(self: Flow[Any], *args: Any, **kwargs: Any) -> Any:
                 # Execute the original method
                 method_output = func(self, *args, **kwargs)
 
@@ -397,11 +400,10 @@ def human_feedback(
         )
         wrapper.__is_flow_method__ = True
 
-        # Make it a router if emit specified
         if emit:
             wrapper.__is_router__ = True
             wrapper.__router_paths__ = list(emit)
 
-        return wrapper  # type: ignore[return-value]
+        return wrapper  # type: ignore[no-any-return]
 
     return decorator

lib/crewai/src/crewai/flow/persistence/base.py

@@ -7,6 +7,7 @@ from typing import TYPE_CHECKING, Any
 
 from pydantic import BaseModel
 
+
 if TYPE_CHECKING:
     from crewai.flow.async_feedback.types import PendingFeedbackContext
 
@@ -103,4 +104,3 @@ class FlowPersistence(ABC):
         Args:
             flow_uuid: Unique identifier for the flow instance
         """
-        pass

lib/crewai/src/crewai/flow/persistence/sqlite.py

@@ -15,6 +15,7 @@ from pydantic import BaseModel
 from crewai.flow.persistence.base import FlowPersistence
 from crewai.utilities.paths import db_storage_path
 
+
 if TYPE_CHECKING:
     from crewai.flow.async_feedback.types import PendingFeedbackContext
 
@@ -176,7 +177,8 @@ class SQLiteFlowPersistence(FlowPersistence):
             row = cursor.fetchone()
 
         if row:
-            return json.loads(row[0])
+            result = json.loads(row[0])
+            return result if isinstance(result, dict) else None
         return None
 
     def save_pending_feedback(
@@ -196,7 +198,6 @@ class SQLiteFlowPersistence(FlowPersistence):
             state_data: Current state data
         """
         # Import here to avoid circular imports
-        from crewai.flow.async_feedback.types import PendingFeedbackContext
 
         # Convert state_data to dict
         if isinstance(state_data, BaseModel):

lib/crewai/src/crewai/hooks/llm_hooks.py

@@ -3,7 +3,12 @@ from __future__ import annotations
 from typing import TYPE_CHECKING, Any, cast
 
 from crewai.events.event_listener import event_listener
-from crewai.hooks.types import AfterLLMCallHookType, BeforeLLMCallHookType
+from crewai.hooks.types import (
+    AfterLLMCallHookCallable,
+    AfterLLMCallHookType,
+    BeforeLLMCallHookCallable,
+    BeforeLLMCallHookType,
+)
 from crewai.utilities.printer import Printer
 
 
@@ -149,12 +154,12 @@ class LLMCallHookContext:
             event_listener.formatter.resume_live_updates()
 
 
-_before_llm_call_hooks: list[BeforeLLMCallHookType] = []
-_after_llm_call_hooks: list[AfterLLMCallHookType] = []
+_before_llm_call_hooks: list[BeforeLLMCallHookType | BeforeLLMCallHookCallable] = []
+_after_llm_call_hooks: list[AfterLLMCallHookType | AfterLLMCallHookCallable] = []
 
 
 def register_before_llm_call_hook(
-    hook: BeforeLLMCallHookType,
+    hook: BeforeLLMCallHookType | BeforeLLMCallHookCallable,
 ) -> None:
     """Register a global before_llm_call hook.
 
@@ -190,7 +195,7 @@ def register_before_llm_call_hook(
 
 
 def register_after_llm_call_hook(
-    hook: AfterLLMCallHookType,
+    hook: AfterLLMCallHookType | AfterLLMCallHookCallable,
 ) -> None:
     """Register a global after_llm_call hook.
 
@@ -217,7 +222,9 @@ def register_after_llm_call_hook(
     _after_llm_call_hooks.append(hook)
 
 
-def get_before_llm_call_hooks() -> list[BeforeLLMCallHookType]:
+def get_before_llm_call_hooks() -> list[
+    BeforeLLMCallHookType | BeforeLLMCallHookCallable
+]:
     """Get all registered global before_llm_call hooks.
 
     Returns:
@@ -226,7 +233,7 @@ def get_before_llm_call_hooks() -> list[BeforeLLMCallHookType]:
     return _before_llm_call_hooks.copy()
 
 
-def get_after_llm_call_hooks() -> list[AfterLLMCallHookType]:
+def get_after_llm_call_hooks() -> list[AfterLLMCallHookType | AfterLLMCallHookCallable]:
     """Get all registered global after_llm_call hooks.
 
     Returns:
@@ -236,7 +243,7 @@ def get_after_llm_call_hooks() -> list[AfterLLMCallHookType]:
 
 
 def unregister_before_llm_call_hook(
-    hook: BeforeLLMCallHookType,
+    hook: BeforeLLMCallHookType | BeforeLLMCallHookCallable,
 ) -> bool:
     """Unregister a specific global before_llm_call hook.
 
@@ -262,7 +269,7 @@ def unregister_before_llm_call_hook(
 
 
 def unregister_after_llm_call_hook(
-    hook: AfterLLMCallHookType,
+    hook: AfterLLMCallHookType | AfterLLMCallHookCallable,
 ) -> bool:
     """Unregister a specific global after_llm_call hook.
 

lib/crewai/src/crewai/hooks/tool_hooks.py

@@ -3,7 +3,12 @@ from __future__ import annotations
 from typing import TYPE_CHECKING, Any
 
 from crewai.events.event_listener import event_listener
-from crewai.hooks.types import AfterToolCallHookType, BeforeToolCallHookType
+from crewai.hooks.types import (
+    AfterToolCallHookCallable,
+    AfterToolCallHookType,
+    BeforeToolCallHookCallable,
+    BeforeToolCallHookType,
+)
 from crewai.utilities.printer import Printer
 
 
@@ -112,12 +117,12 @@ class ToolCallHookContext:
 
 
 # Global hook registries
-_before_tool_call_hooks: list[BeforeToolCallHookType] = []
-_after_tool_call_hooks: list[AfterToolCallHookType] = []
+_before_tool_call_hooks: list[BeforeToolCallHookType | BeforeToolCallHookCallable] = []
+_after_tool_call_hooks: list[AfterToolCallHookType | AfterToolCallHookCallable] = []
 
 
 def register_before_tool_call_hook(
-    hook: BeforeToolCallHookType,
+    hook: BeforeToolCallHookType | BeforeToolCallHookCallable,
 ) -> None:
     """Register a global before_tool_call hook.
 
@@ -154,7 +159,7 @@ def register_before_tool_call_hook(
 
 
 def register_after_tool_call_hook(
-    hook: AfterToolCallHookType,
+    hook: AfterToolCallHookType | AfterToolCallHookCallable,
 ) -> None:
     """Register a global after_tool_call hook.
 
@@ -184,7 +189,9 @@ def register_after_tool_call_hook(
     _after_tool_call_hooks.append(hook)
 
 
-def get_before_tool_call_hooks() -> list[BeforeToolCallHookType]:
+def get_before_tool_call_hooks() -> list[
+    BeforeToolCallHookType | BeforeToolCallHookCallable
+]:
     """Get all registered global before_tool_call hooks.
 
     Returns:
@@ -193,7 +200,9 @@ def get_before_tool_call_hooks() -> list[BeforeToolCallHookType]:
     return _before_tool_call_hooks.copy()
 
 
-def get_after_tool_call_hooks() -> list[AfterToolCallHookType]:
+def get_after_tool_call_hooks() -> list[
+    AfterToolCallHookType | AfterToolCallHookCallable
+]:
     """Get all registered global after_tool_call hooks.
 
     Returns:
@@ -203,7 +212,7 @@ def get_after_tool_call_hooks() -> list[AfterToolCallHookType]:
 
 
 def unregister_before_tool_call_hook(
-    hook: BeforeToolCallHookType,
+    hook: BeforeToolCallHookType | BeforeToolCallHookCallable,
 ) -> bool:
     """Unregister a specific global before_tool_call hook.
 
@@ -229,7 +238,7 @@ def unregister_before_tool_call_hook(
 
 
 def unregister_after_tool_call_hook(
-    hook: AfterToolCallHookType,
+    hook: AfterToolCallHookType | AfterToolCallHookCallable,
 ) -> bool:
     """Unregister a specific global after_tool_call hook.
 

lib/crewai/src/crewai/knowledge/source/utils/__init__.py

@@ -0,0 +1 @@
+"""Knowledge source utilities."""

lib/crewai/src/crewai/knowledge/source/utils/source_helper.py

@@ -0,0 +1,70 @@
+"""Helper utilities for knowledge sources."""
+
+from typing import Any, ClassVar
+
+from crewai.knowledge.source.base_knowledge_source import BaseKnowledgeSource
+from crewai.knowledge.source.csv_knowledge_source import CSVKnowledgeSource
+from crewai.knowledge.source.excel_knowledge_source import ExcelKnowledgeSource
+from crewai.knowledge.source.json_knowledge_source import JSONKnowledgeSource
+from crewai.knowledge.source.pdf_knowledge_source import PDFKnowledgeSource
+from crewai.knowledge.source.text_file_knowledge_source import TextFileKnowledgeSource
+
+
+class SourceHelper:
+    """Helper class for creating and managing knowledge sources."""
+
+    SUPPORTED_FILE_TYPES: ClassVar[list[str]] = [
+        ".csv",
+        ".pdf",
+        ".json",
+        ".txt",
+        ".xlsx",
+        ".xls",
+    ]
+
+    _FILE_TYPE_MAP: ClassVar[dict[str, type[BaseKnowledgeSource]]] = {
+        ".csv": CSVKnowledgeSource,
+        ".pdf": PDFKnowledgeSource,
+        ".json": JSONKnowledgeSource,
+        ".txt": TextFileKnowledgeSource,
+        ".xlsx": ExcelKnowledgeSource,
+        ".xls": ExcelKnowledgeSource,
+    }
+
+    @classmethod
+    def is_supported_file(cls, file_path: str) -> bool:
+        """Check if a file type is supported.
+
+        Args:
+            file_path: Path to the file.
+
+        Returns:
+            True if the file type is supported.
+        """
+        return file_path.lower().endswith(tuple(cls.SUPPORTED_FILE_TYPES))
+
+    @classmethod
+    def get_source(
+        cls, file_path: str, metadata: dict[str, Any] | None = None
+    ) -> BaseKnowledgeSource:
+        """Create appropriate KnowledgeSource based on file extension.
+
+        Args:
+            file_path: Path to the file.
+            metadata: Optional metadata to attach to the source.
+
+        Returns:
+            The appropriate KnowledgeSource instance.
+
+        Raises:
+            ValueError: If the file type is not supported.
+        """
+        if not cls.is_supported_file(file_path):
+            raise ValueError(f"Unsupported file type: {file_path}")
+
+        lower_path = file_path.lower()
+        for ext, source_cls in cls._FILE_TYPE_MAP.items():
+            if lower_path.endswith(ext):
+                return source_cls(file_path=[file_path], metadata=metadata)
+
+        raise ValueError(f"Unsupported file type: {file_path}")

lib/crewai/src/crewai/llm.py

@@ -37,7 +37,7 @@ from crewai.events.types.tool_usage_events import (
     ToolUsageFinishedEvent,
     ToolUsageStartedEvent,
 )
-from crewai.llms.base_llm import BaseLLM
+from crewai.llms.base_llm import BaseLLM, get_current_call_id, llm_call_context
 from crewai.llms.constants import (
     ANTHROPIC_MODELS,
     AZURE_MODELS,
@@ -770,7 +770,7 @@ class LLM(BaseLLM):
                 chunk_content = None
                 response_id = None
 
-                if hasattr(chunk,'id'):
+                if hasattr(chunk, "id"):
                     response_id = chunk.id
 
                 # Safely extract content from various chunk formats
@@ -827,7 +827,7 @@ class LLM(BaseLLM):
                                         available_functions=available_functions,
                                         from_task=from_task,
                                         from_agent=from_agent,
-                                        response_id=response_id
+                                        response_id=response_id,
                                     )
 
                                     if result is not None:
@@ -849,7 +849,8 @@ class LLM(BaseLLM):
                             from_task=from_task,
                             from_agent=from_agent,
                             call_type=LLMCallType.LLM_CALL,
-                            response_id=response_id
+                            response_id=response_id,
+                            call_id=get_current_call_id(),
                         ),
                     )
             # --- 4) Fallback to non-streaming if no content received
@@ -1015,7 +1016,10 @@ class LLM(BaseLLM):
             crewai_event_bus.emit(
                 self,
                 event=LLMCallFailedEvent(
-                    error=str(e), from_task=from_task, from_agent=from_agent
+                    error=str(e),
+                    from_task=from_task,
+                    from_agent=from_agent,
+                    call_id=get_current_call_id(),
                 ),
             )
             raise Exception(f"Failed to get streaming response: {e!s}") from e
@@ -1048,7 +1052,8 @@ class LLM(BaseLLM):
                     from_task=from_task,
                     from_agent=from_agent,
                     call_type=LLMCallType.TOOL_CALL,
-                    response_id=response_id
+                    response_id=response_id,
+                    call_id=get_current_call_id(),
                 ),
             )
 
@@ -1476,7 +1481,8 @@ class LLM(BaseLLM):
                             chunk=chunk_content,
                             from_task=from_task,
                             from_agent=from_agent,
-                            response_id=response_id
+                            response_id=response_id,
+                            call_id=get_current_call_id(),
                         ),
                     )
 
@@ -1619,7 +1625,12 @@ class LLM(BaseLLM):
                 logging.error(f"Error executing function '{function_name}': {e}")
                 crewai_event_bus.emit(
                     self,
-                    event=LLMCallFailedEvent(error=f"Tool execution error: {e!s}"),
+                    event=LLMCallFailedEvent(
+                        error=f"Tool execution error: {e!s}",
+                        from_task=from_task,
+                        from_agent=from_agent,
+                        call_id=get_current_call_id(),
+                    ),
                 )
                 crewai_event_bus.emit(
                     self,
@@ -1669,108 +1680,117 @@ class LLM(BaseLLM):
             ValueError: If response format is not supported
             LLMContextLengthExceededError: If input exceeds model's context limit
         """
-        crewai_event_bus.emit(
-            self,
-            event=LLMCallStartedEvent(
-                messages=messages,
-                tools=tools,
-                callbacks=callbacks,
-                available_functions=available_functions,
-                from_task=from_task,
-                from_agent=from_agent,
-                model=self.model,
-            ),
-        )
+        with llm_call_context() as call_id:
+            crewai_event_bus.emit(
+                self,
+                event=LLMCallStartedEvent(
+                    messages=messages,
+                    tools=tools,
+                    callbacks=callbacks,
+                    available_functions=available_functions,
+                    from_task=from_task,
+                    from_agent=from_agent,
+                    model=self.model,
+                    call_id=call_id,
+                ),
+            )
 
-        # --- 2) Validate parameters before proceeding with the call
-        self._validate_call_params()
+            # --- 2) Validate parameters before proceeding with the call
+            self._validate_call_params()
 
-        # --- 3) Convert string messages to proper format if needed
-        if isinstance(messages, str):
-            messages = [{"role": "user", "content": messages}]
-        # --- 4) Handle O1 model special case (system messages not supported)
-        if "o1" in self.model.lower():
-            for message in messages:
-                if message.get("role") == "system":
-                    msg_role: Literal["assistant"] = "assistant"
-                    message["role"] = msg_role
+            # --- 3) Convert string messages to proper format if needed
+            if isinstance(messages, str):
+                messages = [{"role": "user", "content": messages}]
+            # --- 4) Handle O1 model special case (system messages not supported)
+            if "o1" in self.model.lower():
+                for message in messages:
+                    if message.get("role") == "system":
+                        msg_role: Literal["assistant"] = "assistant"
+                        message["role"] = msg_role
 
-        if not self._invoke_before_llm_call_hooks(messages, from_agent):
-            raise ValueError("LLM call blocked by before_llm_call hook")
+            if not self._invoke_before_llm_call_hooks(messages, from_agent):
+                raise ValueError("LLM call blocked by before_llm_call hook")
 
-        # --- 5) Set up callbacks if provided
-        with suppress_warnings():
-            if callbacks and len(callbacks) > 0:
-                self.set_callbacks(callbacks)
-            try:
-                # --- 6) Prepare parameters for the completion call
-                params = self._prepare_completion_params(messages, tools)
-                # --- 7) Make the completion call and handle response
-                if self.stream:
-                    result = self._handle_streaming_response(
-                        params=params,
-                        callbacks=callbacks,
-                        available_functions=available_functions,
-                        from_task=from_task,
-                        from_agent=from_agent,
-                        response_model=response_model,
-                    )
-                else:
-                    result = self._handle_non_streaming_response(
-                        params=params,
-                        callbacks=callbacks,
-                        available_functions=available_functions,
-                        from_task=from_task,
-                        from_agent=from_agent,
-                        response_model=response_model,
-                    )
-
-                if isinstance(result, str):
-                    result = self._invoke_after_llm_call_hooks(
-                        messages, result, from_agent
-                    )
-
-                return result
-            except LLMContextLengthExceededError:
-                # Re-raise LLMContextLengthExceededError as it should be handled
-                # by the CrewAgentExecutor._invoke_loop method, which can then decide
-                # whether to summarize the content or abort based on the respect_context_window flag
-                raise
-            except Exception as e:
-                unsupported_stop = "Unsupported parameter" in str(
-                    e
-                ) and "'stop'" in str(e)
-
-                if unsupported_stop:
-                    if (
-                        "additional_drop_params" in self.additional_params
-                        and isinstance(
-                            self.additional_params["additional_drop_params"], list
+            # --- 5) Set up callbacks if provided
+            with suppress_warnings():
+                if callbacks and len(callbacks) > 0:
+                    self.set_callbacks(callbacks)
+                try:
+                    # --- 6) Prepare parameters for the completion call
+                    params = self._prepare_completion_params(messages, tools)
+                    # --- 7) Make the completion call and handle response
+                    if self.stream:
+                        result = self._handle_streaming_response(
+                            params=params,
+                            callbacks=callbacks,
+                            available_functions=available_functions,
+                            from_task=from_task,
+                            from_agent=from_agent,
+                            response_model=response_model,
                         )
-                    ):
-                        self.additional_params["additional_drop_params"].append("stop")
                     else:
-                        self.additional_params = {"additional_drop_params": ["stop"]}
+                        result = self._handle_non_streaming_response(
+                            params=params,
+                            callbacks=callbacks,
+                            available_functions=available_functions,
+                            from_task=from_task,
+                            from_agent=from_agent,
+                            response_model=response_model,
+                        )
 
-                    logging.info("Retrying LLM call without the unsupported 'stop'")
+                    if isinstance(result, str):
+                        result = self._invoke_after_llm_call_hooks(
+                            messages, result, from_agent
+                        )
 
-                    return self.call(
-                        messages,
-                        tools=tools,
-                        callbacks=callbacks,
-                        available_functions=available_functions,
-                        from_task=from_task,
-                        from_agent=from_agent,
-                        response_model=response_model,
+                    return result
+                except LLMContextLengthExceededError:
+                    # Re-raise LLMContextLengthExceededError as it should be handled
+                    # by the CrewAgentExecutor._invoke_loop method, which can then decide
+                    # whether to summarize the content or abort based on the respect_context_window flag
+                    raise
+                except Exception as e:
+                    unsupported_stop = "Unsupported parameter" in str(
+                        e
+                    ) and "'stop'" in str(e)
+
+                    if unsupported_stop:
+                        if (
+                            "additional_drop_params" in self.additional_params
+                            and isinstance(
+                                self.additional_params["additional_drop_params"], list
+                            )
+                        ):
+                            self.additional_params["additional_drop_params"].append(
+                                "stop"
+                            )
+                        else:
+                            self.additional_params = {
+                                "additional_drop_params": ["stop"]
+                            }
+
+                        logging.info("Retrying LLM call without the unsupported 'stop'")
+
+                        return self.call(
+                            messages,
+                            tools=tools,
+                            callbacks=callbacks,
+                            available_functions=available_functions,
+                            from_task=from_task,
+                            from_agent=from_agent,
+                            response_model=response_model,
+                        )
+
+                    crewai_event_bus.emit(
+                        self,
+                        event=LLMCallFailedEvent(
+                            error=str(e),
+                            from_task=from_task,
+                            from_agent=from_agent,
+                            call_id=get_current_call_id(),
+                        ),
                     )
-
-                crewai_event_bus.emit(
-                    self,
-                    event=LLMCallFailedEvent(
-                        error=str(e), from_task=from_task, from_agent=from_agent
-                    ),
-                )
-                raise
+                    raise
 
     async def acall(
         self,
@@ -1808,43 +1828,54 @@ class LLM(BaseLLM):
             ValueError: If response format is not supported
             LLMContextLengthExceededError: If input exceeds model's context limit
         """
-        crewai_event_bus.emit(
-            self,
-            event=LLMCallStartedEvent(
-                messages=messages,
-                tools=tools,
-                callbacks=callbacks,
-                available_functions=available_functions,
-                from_task=from_task,
-                from_agent=from_agent,
-                model=self.model,
-            ),
-        )
+        with llm_call_context() as call_id:
+            crewai_event_bus.emit(
+                self,
+                event=LLMCallStartedEvent(
+                    messages=messages,
+                    tools=tools,
+                    callbacks=callbacks,
+                    available_functions=available_functions,
+                    from_task=from_task,
+                    from_agent=from_agent,
+                    model=self.model,
+                    call_id=call_id,
+                ),
+            )
 
-        self._validate_call_params()
+            self._validate_call_params()
 
-        if isinstance(messages, str):
-            messages = [{"role": "user", "content": messages}]
+            if isinstance(messages, str):
+                messages = [{"role": "user", "content": messages}]
 
-        # Process file attachments asynchronously before preparing params
-        messages = await self._aprocess_message_files(messages)
+            # Process file attachments asynchronously before preparing params
+            messages = await self._aprocess_message_files(messages)
 
-        if "o1" in self.model.lower():
-            for message in messages:
-                if message.get("role") == "system":
-                    msg_role: Literal["assistant"] = "assistant"
-                    message["role"] = msg_role
+            if "o1" in self.model.lower():
+                for message in messages:
+                    if message.get("role") == "system":
+                        msg_role: Literal["assistant"] = "assistant"
+                        message["role"] = msg_role
 
-        with suppress_warnings():
-            if callbacks and len(callbacks) > 0:
-                self.set_callbacks(callbacks)
-            try:
-                params = self._prepare_completion_params(
-                    messages, tools, skip_file_processing=True
-                )
+            with suppress_warnings():
+                if callbacks and len(callbacks) > 0:
+                    self.set_callbacks(callbacks)
+                try:
+                    params = self._prepare_completion_params(
+                        messages, tools, skip_file_processing=True
+                    )
 
-                if self.stream:
-                    return await self._ahandle_streaming_response(
+                    if self.stream:
+                        return await self._ahandle_streaming_response(
+                            params=params,
+                            callbacks=callbacks,
+                            available_functions=available_functions,
+                            from_task=from_task,
+                            from_agent=from_agent,
+                            response_model=response_model,
+                        )
+
+                    return await self._ahandle_non_streaming_response(
                         params=params,
                         callbacks=callbacks,
                         available_functions=available_functions,
@@ -1852,52 +1883,50 @@ class LLM(BaseLLM):
                         from_agent=from_agent,
                         response_model=response_model,
                     )
+                except LLMContextLengthExceededError:
+                    raise
+                except Exception as e:
+                    unsupported_stop = "Unsupported parameter" in str(
+                        e
+                    ) and "'stop'" in str(e)
 
-                return await self._ahandle_non_streaming_response(
-                    params=params,
-                    callbacks=callbacks,
-                    available_functions=available_functions,
-                    from_task=from_task,
-                    from_agent=from_agent,
-                    response_model=response_model,
-                )
-            except LLMContextLengthExceededError:
-                raise
-            except Exception as e:
-                unsupported_stop = "Unsupported parameter" in str(
-                    e
-                ) and "'stop'" in str(e)
+                    if unsupported_stop:
+                        if (
+                            "additional_drop_params" in self.additional_params
+                            and isinstance(
+                                self.additional_params["additional_drop_params"], list
+                            )
+                        ):
+                            self.additional_params["additional_drop_params"].append(
+                                "stop"
+                            )
+                        else:
+                            self.additional_params = {
+                                "additional_drop_params": ["stop"]
+                            }
 
-                if unsupported_stop:
-                    if (
-                        "additional_drop_params" in self.additional_params
-                        and isinstance(
-                            self.additional_params["additional_drop_params"], list
+                        logging.info("Retrying LLM call without the unsupported 'stop'")
+
+                        return await self.acall(
+                            messages,
+                            tools=tools,
+                            callbacks=callbacks,
+                            available_functions=available_functions,
+                            from_task=from_task,
+                            from_agent=from_agent,
+                            response_model=response_model,
                         )
-                    ):
-                        self.additional_params["additional_drop_params"].append("stop")
-                    else:
-                        self.additional_params = {"additional_drop_params": ["stop"]}
 
-                    logging.info("Retrying LLM call without the unsupported 'stop'")
-
-                    return await self.acall(
-                        messages,
-                        tools=tools,
-                        callbacks=callbacks,
-                        available_functions=available_functions,
-                        from_task=from_task,
-                        from_agent=from_agent,
-                        response_model=response_model,
+                    crewai_event_bus.emit(
+                        self,
+                        event=LLMCallFailedEvent(
+                            error=str(e),
+                            from_task=from_task,
+                            from_agent=from_agent,
+                            call_id=get_current_call_id(),
+                        ),
                     )
-
-                crewai_event_bus.emit(
-                    self,
-                    event=LLMCallFailedEvent(
-                        error=str(e), from_task=from_task, from_agent=from_agent
-                    ),
-                )
-                raise
+                    raise
 
     def _handle_emit_call_events(
         self,
@@ -1925,6 +1954,7 @@ class LLM(BaseLLM):
                 from_task=from_task,
                 from_agent=from_agent,
                 model=self.model,
+                call_id=get_current_call_id(),
             ),
         )
 

lib/crewai/src/crewai/llms/base_llm.py

@@ -7,11 +7,15 @@ in CrewAI, including common functionality for native SDK implementations.
 from __future__ import annotations
 
 from abc import ABC, abstractmethod
+from collections.abc import Generator
+from contextlib import contextmanager
+import contextvars
 from datetime import datetime
 import json
 import logging
 import re
 from typing import TYPE_CHECKING, Any, Final
+import uuid
 
 from pydantic import BaseModel
 
@@ -50,6 +54,32 @@ DEFAULT_CONTEXT_WINDOW_SIZE: Final[int] = 4096
 DEFAULT_SUPPORTS_STOP_WORDS: Final[bool] = True
 _JSON_EXTRACTION_PATTERN: Final[re.Pattern[str]] = re.compile(r"\{.*}", re.DOTALL)
 
+_current_call_id: contextvars.ContextVar[str | None] = contextvars.ContextVar(
+    "_current_call_id", default=None
+)
+
+
+@contextmanager
+def llm_call_context() -> Generator[str, None, None]:
+    """Context manager that establishes an LLM call scope with a unique call_id."""
+    call_id = str(uuid.uuid4())
+    token = _current_call_id.set(call_id)
+    try:
+        yield call_id
+    finally:
+        _current_call_id.reset(token)
+
+
+def get_current_call_id() -> str:
+    """Get current call_id from context"""
+    call_id = _current_call_id.get()
+    if call_id is None:
+        logging.warning(
+            "LLM event emitted outside call context - generating fallback call_id"
+        )
+        return str(uuid.uuid4())
+    return call_id
+
 
 class BaseLLM(ABC):
     """Abstract base class for LLM implementations.
@@ -351,6 +381,7 @@ class BaseLLM(ABC):
                 from_task=from_task,
                 from_agent=from_agent,
                 model=self.model,
+                call_id=get_current_call_id(),
             ),
         )
 
@@ -374,6 +405,7 @@ class BaseLLM(ABC):
                 from_task=from_task,
                 from_agent=from_agent,
                 model=self.model,
+                call_id=get_current_call_id(),
             ),
         )
 
@@ -394,6 +426,7 @@ class BaseLLM(ABC):
                 from_task=from_task,
                 from_agent=from_agent,
                 model=self.model,
+                call_id=get_current_call_id(),
             ),
         )
 
@@ -428,6 +461,7 @@ class BaseLLM(ABC):
                 from_agent=from_agent,
                 call_type=call_type,
                 response_id=response_id,
+                call_id=get_current_call_id(),
             ),
         )
 

lib/crewai/src/crewai/llms/providers/anthropic/completion.py

@@ -8,7 +8,7 @@ from typing import TYPE_CHECKING, Any, Final, Literal, TypeGuard, cast
 from pydantic import BaseModel
 
 from crewai.events.types.llm_events import LLMCallType
-from crewai.llms.base_llm import BaseLLM
+from crewai.llms.base_llm import BaseLLM, llm_call_context
 from crewai.llms.hooks.transport import AsyncHTTPTransport, HTTPTransport
 from crewai.utilities.agent_utils import is_context_length_exceeded
 from crewai.utilities.exceptions.context_window_exceeding_exception import (
@@ -266,35 +266,46 @@ class AnthropicCompletion(BaseLLM):
         Returns:
             Chat completion response or tool call result
         """
-        try:
-            # Emit call started event
-            self._emit_call_started_event(
-                messages=messages,
-                tools=tools,
-                callbacks=callbacks,
-                available_functions=available_functions,
-                from_task=from_task,
-                from_agent=from_agent,
-            )
+        with llm_call_context():
+            try:
+                # Emit call started event
+                self._emit_call_started_event(
+                    messages=messages,
+                    tools=tools,
+                    callbacks=callbacks,
+                    available_functions=available_functions,
+                    from_task=from_task,
+                    from_agent=from_agent,
+                )
 
-            # Format messages for Anthropic
-            formatted_messages, system_message = self._format_messages_for_anthropic(
-                messages
-            )
+                # Format messages for Anthropic
+                formatted_messages, system_message = (
+                    self._format_messages_for_anthropic(messages)
+                )
 
-            if not self._invoke_before_llm_call_hooks(formatted_messages, from_agent):
-                raise ValueError("LLM call blocked by before_llm_call hook")
+                if not self._invoke_before_llm_call_hooks(
+                    formatted_messages, from_agent
+                ):
+                    raise ValueError("LLM call blocked by before_llm_call hook")
 
-            # Prepare completion parameters
-            completion_params = self._prepare_completion_params(
-                formatted_messages, system_message, tools
-            )
+                # Prepare completion parameters
+                completion_params = self._prepare_completion_params(
+                    formatted_messages, system_message, tools, available_functions
+                )
 
-            effective_response_model = response_model or self.response_format
+                effective_response_model = response_model or self.response_format
 
-            # Handle streaming vs non-streaming
-            if self.stream:
-                return self._handle_streaming_completion(
+                # Handle streaming vs non-streaming
+                if self.stream:
+                    return self._handle_streaming_completion(
+                        completion_params,
+                        available_functions,
+                        from_task,
+                        from_agent,
+                        effective_response_model,
+                    )
+
+                return self._handle_completion(
                     completion_params,
                     available_functions,
                     from_task,
@@ -302,21 +313,13 @@ class AnthropicCompletion(BaseLLM):
                     effective_response_model,
                 )
 
-            return self._handle_completion(
-                completion_params,
-                available_functions,
-                from_task,
-                from_agent,
-                effective_response_model,
-            )
-
-        except Exception as e:
-            error_msg = f"Anthropic API call failed: {e!s}"
-            logging.error(error_msg)
-            self._emit_call_failed_event(
-                error=error_msg, from_task=from_task, from_agent=from_agent
-            )
-            raise
+            except Exception as e:
+                error_msg = f"Anthropic API call failed: {e!s}"
+                logging.error(error_msg)
+                self._emit_call_failed_event(
+                    error=error_msg, from_task=from_task, from_agent=from_agent
+                )
+                raise
 
     async def acall(
         self,
@@ -342,28 +345,37 @@ class AnthropicCompletion(BaseLLM):
         Returns:
             Chat completion response or tool call result
         """
-        try:
-            self._emit_call_started_event(
-                messages=messages,
-                tools=tools,
-                callbacks=callbacks,
-                available_functions=available_functions,
-                from_task=from_task,
-                from_agent=from_agent,
-            )
+        with llm_call_context():
+            try:
+                self._emit_call_started_event(
+                    messages=messages,
+                    tools=tools,
+                    callbacks=callbacks,
+                    available_functions=available_functions,
+                    from_task=from_task,
+                    from_agent=from_agent,
+                )
 
-            formatted_messages, system_message = self._format_messages_for_anthropic(
-                messages
-            )
+                formatted_messages, system_message = (
+                    self._format_messages_for_anthropic(messages)
+                )
 
-            completion_params = self._prepare_completion_params(
-                formatted_messages, system_message, tools
-            )
+                completion_params = self._prepare_completion_params(
+                    formatted_messages, system_message, tools, available_functions
+                )
 
-            effective_response_model = response_model or self.response_format
+                effective_response_model = response_model or self.response_format
 
-            if self.stream:
-                return await self._ahandle_streaming_completion(
+                if self.stream:
+                    return await self._ahandle_streaming_completion(
+                        completion_params,
+                        available_functions,
+                        from_task,
+                        from_agent,
+                        effective_response_model,
+                    )
+
+                return await self._ahandle_completion(
                     completion_params,
                     available_functions,
                     from_task,
@@ -371,27 +383,20 @@ class AnthropicCompletion(BaseLLM):
                     effective_response_model,
                 )
 
-            return await self._ahandle_completion(
-                completion_params,
-                available_functions,
-                from_task,
-                from_agent,
-                effective_response_model,
-            )
-
-        except Exception as e:
-            error_msg = f"Anthropic API call failed: {e!s}"
-            logging.error(error_msg)
-            self._emit_call_failed_event(
-                error=error_msg, from_task=from_task, from_agent=from_agent
-            )
-            raise
+            except Exception as e:
+                error_msg = f"Anthropic API call failed: {e!s}"
+                logging.error(error_msg)
+                self._emit_call_failed_event(
+                    error=error_msg, from_task=from_task, from_agent=from_agent
+                )
+                raise
 
     def _prepare_completion_params(
         self,
         messages: list[LLMMessage],
         system_message: str | None = None,
         tools: list[dict[str, Any]] | None = None,
+        available_functions: dict[str, Any] | None = None,
     ) -> dict[str, Any]:
         """Prepare parameters for Anthropic messages API.
 
@@ -399,6 +404,8 @@ class AnthropicCompletion(BaseLLM):
             messages: Formatted messages for Anthropic
             system_message: Extracted system message
             tools: Tool definitions
+            available_functions: Available functions for tool calling. When provided
+                with a single tool, tool_choice is automatically set to force tool use.
 
         Returns:
             Parameters dictionary for Anthropic API
@@ -424,7 +431,13 @@ class AnthropicCompletion(BaseLLM):
 
         # Handle tools for Claude 3+
         if tools and self.supports_tools:
-            params["tools"] = self._convert_tools_for_interference(tools)
+            converted_tools = self._convert_tools_for_interference(tools)
+            params["tools"] = converted_tools
+
+            if available_functions and len(converted_tools) == 1:
+                tool_name = converted_tools[0].get("name")
+                if tool_name and tool_name in available_functions:
+                    params["tool_choice"] = {"type": "tool", "name": tool_name}
 
         if self.thinking:
             if isinstance(self.thinking, AnthropicThinkingConfig):
@@ -726,15 +739,11 @@ class AnthropicCompletion(BaseLLM):
                     )
                     return list(tool_uses)
 
-                # Handle tool use conversation flow internally
-                return self._handle_tool_use_conversation(
-                    response,
-                    tool_uses,
-                    params,
-                    available_functions,
-                    from_task,
-                    from_agent,
+                result = self._execute_first_tool(
+                    tool_uses, available_functions, from_task, from_agent
                 )
+                if result is not None:
+                    return result
 
         content = ""
         thinking_blocks: list[ThinkingBlock] = []
@@ -935,14 +944,12 @@ class AnthropicCompletion(BaseLLM):
                 if not available_functions:
                     return list(tool_uses)
 
-                return self._handle_tool_use_conversation(
-                    final_message,
-                    tool_uses,
-                    params,
-                    available_functions,
-                    from_task,
-                    from_agent,
+                # Execute first tool and return result directly
+                result = self._execute_first_tool(
+                    tool_uses, available_functions, from_task, from_agent
                 )
+                if result is not None:
+                    return result
 
         full_response = self._apply_stop_words(full_response)
 
@@ -1001,6 +1008,41 @@ class AnthropicCompletion(BaseLLM):
 
         return tool_results
 
+    def _execute_first_tool(
+        self,
+        tool_uses: list[ToolUseBlock | BetaToolUseBlock],
+        available_functions: dict[str, Any],
+        from_task: Any | None = None,
+        from_agent: Any | None = None,
+    ) -> Any | None:
+        """Execute the first tool from the tool_uses list and return its result.
+
+        This is used when available_functions is provided, to directly execute
+        the tool and return its result (matching OpenAI behavior for use cases
+        like reasoning_handler).
+
+        Args:
+            tool_uses: List of tool use blocks from Claude's response
+            available_functions: Available functions for tool calling
+            from_task: Task that initiated the call
+            from_agent: Agent that initiated the call
+
+        Returns:
+            The result of the first tool execution, or None if execution failed
+        """
+        tool_use = tool_uses[0]
+        function_name = tool_use.name
+        function_args = cast(dict[str, Any], tool_use.input)
+
+        return self._handle_tool_execution(
+            function_name=function_name,
+            function_args=function_args,
+            available_functions=available_functions,
+            from_task=from_task,
+            from_agent=from_agent,
+        )
+
+    # TODO: we drop this
     def _handle_tool_use_conversation(
         self,
         initial_response: Message | BetaMessage,
@@ -1216,14 +1258,11 @@ class AnthropicCompletion(BaseLLM):
                     )
                     return list(tool_uses)
 
-                return await self._ahandle_tool_use_conversation(
-                    response,
-                    tool_uses,
-                    params,
-                    available_functions,
-                    from_task,
-                    from_agent,
+                result = self._execute_first_tool(
+                    tool_uses, available_functions, from_task, from_agent
                 )
+                if result is not None:
+                    return result
 
         content = ""
         if response.content:
@@ -1404,14 +1443,11 @@ class AnthropicCompletion(BaseLLM):
                 if not available_functions:
                     return list(tool_uses)
 
-                return await self._ahandle_tool_use_conversation(
-                    final_message,
-                    tool_uses,
-                    params,
-                    available_functions,
-                    from_task,
-                    from_agent,
+                result = self._execute_first_tool(
+                    tool_uses, available_functions, from_task, from_agent
                 )
+                if result is not None:
+                    return result
 
         full_response = self._apply_stop_words(full_response)
 

lib/crewai/src/crewai/llms/providers/azure/completion.py

@@ -43,7 +43,7 @@ try:
     )
 
     from crewai.events.types.llm_events import LLMCallType
-    from crewai.llms.base_llm import BaseLLM
+    from crewai.llms.base_llm import BaseLLM, llm_call_context
 
 except ImportError:
     raise ImportError(
@@ -293,32 +293,44 @@ class AzureCompletion(BaseLLM):
         Returns:
             Chat completion response or tool call result
         """
-        try:
-            # Emit call started event
-            self._emit_call_started_event(
-                messages=messages,
-                tools=tools,
-                callbacks=callbacks,
-                available_functions=available_functions,
-                from_task=from_task,
-                from_agent=from_agent,
-            )
-            effective_response_model = response_model or self.response_format
+        with llm_call_context():
+            try:
+                # Emit call started event
+                self._emit_call_started_event(
+                    messages=messages,
+                    tools=tools,
+                    callbacks=callbacks,
+                    available_functions=available_functions,
+                    from_task=from_task,
+                    from_agent=from_agent,
+                )
 
-            # Format messages for Azure
-            formatted_messages = self._format_messages_for_azure(messages)
+                effective_response_model = response_model or self.response_format
 
-            if not self._invoke_before_llm_call_hooks(formatted_messages, from_agent):
-                raise ValueError("LLM call blocked by before_llm_call hook")
+                # Format messages for Azure
+                formatted_messages = self._format_messages_for_azure(messages)
 
-            # Prepare completion parameters
-            completion_params = self._prepare_completion_params(
-                formatted_messages, tools, effective_response_model
-            )
+                if not self._invoke_before_llm_call_hooks(
+                    formatted_messages, from_agent
+                ):
+                    raise ValueError("LLM call blocked by before_llm_call hook")
 
-            # Handle streaming vs non-streaming
-            if self.stream:
-                return self._handle_streaming_completion(
+                # Prepare completion parameters
+                completion_params = self._prepare_completion_params(
+                    formatted_messages, tools, effective_response_model
+                )
+
+                # Handle streaming vs non-streaming
+                if self.stream:
+                    return self._handle_streaming_completion(
+                        completion_params,
+                        available_functions,
+                        from_task,
+                        from_agent,
+                        effective_response_model,
+                    )
+
+                return self._handle_completion(
                     completion_params,
                     available_functions,
                     from_task,
@@ -326,16 +338,8 @@ class AzureCompletion(BaseLLM):
                     effective_response_model,
                 )
 
-            return self._handle_completion(
-                completion_params,
-                available_functions,
-                from_task,
-                from_agent,
-                effective_response_model,
-            )
-
-        except Exception as e:
-            return self._handle_api_error(e, from_task, from_agent)  # type: ignore[func-returns-value]
+            except Exception as e:
+                return self._handle_api_error(e, from_task, from_agent)  # type: ignore[func-returns-value]
 
     async def acall(  # type: ignore[return]
         self,
@@ -361,25 +365,35 @@ class AzureCompletion(BaseLLM):
         Returns:
             Chat completion response or tool call result
         """
-        try:
-            self._emit_call_started_event(
-                messages=messages,
-                tools=tools,
-                callbacks=callbacks,
-                available_functions=available_functions,
-                from_task=from_task,
-                from_agent=from_agent,
-            )
-            effective_response_model = response_model or self.response_format
+        with llm_call_context():
+            try:
+                self._emit_call_started_event(
+                    messages=messages,
+                    tools=tools,
+                    callbacks=callbacks,
+                    available_functions=available_functions,
+                    from_task=from_task,
+                    from_agent=from_agent,
+                )
 
-            formatted_messages = self._format_messages_for_azure(messages)
+                effective_response_model = response_model or self.response_format
 
-            completion_params = self._prepare_completion_params(
-                formatted_messages, tools, effective_response_model
-            )
+                formatted_messages = self._format_messages_for_azure(messages)
 
-            if self.stream:
-                return await self._ahandle_streaming_completion(
+                completion_params = self._prepare_completion_params(
+                    formatted_messages, tools, effective_response_model
+                )
+
+                if self.stream:
+                    return await self._ahandle_streaming_completion(
+                        completion_params,
+                        available_functions,
+                        from_task,
+                        from_agent,
+                        effective_response_model,
+                    )
+
+                return await self._ahandle_completion(
                     completion_params,
                     available_functions,
                     from_task,
@@ -387,16 +401,8 @@ class AzureCompletion(BaseLLM):
                     effective_response_model,
                 )
 
-            return await self._ahandle_completion(
-                completion_params,
-                available_functions,
-                from_task,
-                from_agent,
-                effective_response_model,
-            )
-
-        except Exception as e:
-            self._handle_api_error(e, from_task, from_agent)
+            except Exception as e:
+                self._handle_api_error(e, from_task, from_agent)
 
     def _prepare_completion_params(
         self,

lib/crewai/src/crewai/llms/providers/bedrock/completion.py

@@ -11,7 +11,7 @@ from pydantic import BaseModel
 from typing_extensions import Required
 
 from crewai.events.types.llm_events import LLMCallType
-from crewai.llms.base_llm import BaseLLM
+from crewai.llms.base_llm import BaseLLM, llm_call_context
 from crewai.utilities.agent_utils import is_context_length_exceeded
 from crewai.utilities.exceptions.context_window_exceeding_exception import (
     LLMContextLengthExceededError,
@@ -378,77 +378,90 @@ class BedrockCompletion(BaseLLM):
         """Call AWS Bedrock Converse API."""
         effective_response_model = response_model or self.response_format
 
-        try:
-            # Emit call started event
-            self._emit_call_started_event(
-                messages=messages,
-                tools=tools,
-                callbacks=callbacks,
-                available_functions=available_functions,
-                from_task=from_task,
-                from_agent=from_agent,
-            )
-
-            # Format messages for Converse API
-            formatted_messages, system_message = self._format_messages_for_converse(
-                messages
-            )
-
-            if not self._invoke_before_llm_call_hooks(formatted_messages, from_agent):
-                raise ValueError("LLM call blocked by before_llm_call hook")
-
-            # Prepare request body
-            body: BedrockConverseRequestBody = {
-                "inferenceConfig": self._get_inference_config(),
-            }
-
-            # Add system message if present
-            if system_message:
-                body["system"] = cast(
-                    "list[SystemContentBlockTypeDef]",
-                    cast(object, [{"text": system_message}]),
+        with llm_call_context():
+            try:
+                # Emit call started event
+                self._emit_call_started_event(
+                    messages=messages,
+                    tools=tools,
+                    callbacks=callbacks,
+                    available_functions=available_functions,
+                    from_task=from_task,
+                    from_agent=from_agent,
                 )
 
-            # Add tool config if present or if messages contain tool content
-            # Bedrock requires toolConfig when messages have toolUse/toolResult
-            if tools:
-                tool_config: ToolConfigurationTypeDef = {
-                    "tools": cast(
-                        "Sequence[ToolTypeDef]",
-                        cast(object, self._format_tools_for_converse(tools)),
-                    )
+                # Format messages for Converse API
+                formatted_messages, system_message = self._format_messages_for_converse(
+                    messages
+                )
+
+                if not self._invoke_before_llm_call_hooks(
+                    formatted_messages, from_agent
+                ):
+                    raise ValueError("LLM call blocked by before_llm_call hook")
+
+                # Prepare request body
+                body: BedrockConverseRequestBody = {
+                    "inferenceConfig": self._get_inference_config(),
                 }
-                body["toolConfig"] = tool_config
-            elif self._messages_contain_tool_content(formatted_messages):
-                # Create minimal toolConfig from tool history in messages
-                tools_from_history = self._extract_tools_from_message_history(
-                    formatted_messages
-                )
-                if tools_from_history:
-                    body["toolConfig"] = cast(
-                        "ToolConfigurationTypeDef",
-                        cast(object, {"tools": tools_from_history}),
+
+                # Add system message if present
+                if system_message:
+                    body["system"] = cast(
+                        "list[SystemContentBlockTypeDef]",
+                        cast(object, [{"text": system_message}]),
                     )
 
-            # Add optional advanced features if configured
-            if self.guardrail_config:
-                guardrail_config: GuardrailConfigurationTypeDef = cast(
-                    "GuardrailConfigurationTypeDef", cast(object, self.guardrail_config)
-                )
-                body["guardrailConfig"] = guardrail_config
+                # Add tool config if present or if messages contain tool content
+                # Bedrock requires toolConfig when messages have toolUse/toolResult
+                if tools:
+                    tool_config: ToolConfigurationTypeDef = {
+                        "tools": cast(
+                            "Sequence[ToolTypeDef]",
+                            cast(object, self._format_tools_for_converse(tools)),
+                        )
+                    }
+                    body["toolConfig"] = tool_config
+                elif self._messages_contain_tool_content(formatted_messages):
+                    # Create minimal toolConfig from tool history in messages
+                    tools_from_history = self._extract_tools_from_message_history(
+                        formatted_messages
+                    )
+                    if tools_from_history:
+                        body["toolConfig"] = cast(
+                            "ToolConfigurationTypeDef",
+                            cast(object, {"tools": tools_from_history}),
+                        )
 
-            if self.additional_model_request_fields:
-                body["additionalModelRequestFields"] = (
-                    self.additional_model_request_fields
-                )
+                # Add optional advanced features if configured
+                if self.guardrail_config:
+                    guardrail_config: GuardrailConfigurationTypeDef = cast(
+                        "GuardrailConfigurationTypeDef",
+                        cast(object, self.guardrail_config),
+                    )
+                    body["guardrailConfig"] = guardrail_config
 
-            if self.additional_model_response_field_paths:
-                body["additionalModelResponseFieldPaths"] = (
-                    self.additional_model_response_field_paths
-                )
+                if self.additional_model_request_fields:
+                    body["additionalModelRequestFields"] = (
+                        self.additional_model_request_fields
+                    )
 
-            if self.stream:
-                return self._handle_streaming_converse(
+                if self.additional_model_response_field_paths:
+                    body["additionalModelResponseFieldPaths"] = (
+                        self.additional_model_response_field_paths
+                    )
+
+                if self.stream:
+                    return self._handle_streaming_converse(
+                        formatted_messages,
+                        body,
+                        available_functions,
+                        from_task,
+                        from_agent,
+                        effective_response_model,
+                    )
+
+                return self._handle_converse(
                     formatted_messages,
                     body,
                     available_functions,
@@ -457,26 +470,17 @@ class BedrockCompletion(BaseLLM):
                     effective_response_model,
                 )
 
-            return self._handle_converse(
-                formatted_messages,
-                body,
-                available_functions,
-                from_task,
-                from_agent,
-                effective_response_model,
-            )
+            except Exception as e:
+                if is_context_length_exceeded(e):
+                    logging.error(f"Context window exceeded: {e}")
+                    raise LLMContextLengthExceededError(str(e)) from e
 
-        except Exception as e:
-            if is_context_length_exceeded(e):
-                logging.error(f"Context window exceeded: {e}")
-                raise LLMContextLengthExceededError(str(e)) from e
-
-            error_msg = f"AWS Bedrock API call failed: {e!s}"
-            logging.error(error_msg)
-            self._emit_call_failed_event(
-                error=error_msg, from_task=from_task, from_agent=from_agent
-            )
-            raise
+                error_msg = f"AWS Bedrock API call failed: {e!s}"
+                logging.error(error_msg)
+                self._emit_call_failed_event(
+                    error=error_msg, from_task=from_task, from_agent=from_agent
+                )
+                raise
 
     async def acall(
         self,
@@ -514,69 +518,80 @@ class BedrockCompletion(BaseLLM):
                 'Install with: uv add "crewai[bedrock-async]"'
             )
 
-        try:
-            self._emit_call_started_event(
-                messages=messages,
-                tools=tools,
-                callbacks=callbacks,
-                available_functions=available_functions,
-                from_task=from_task,
-                from_agent=from_agent,
-            )
-
-            formatted_messages, system_message = self._format_messages_for_converse(
-                messages
-            )
-
-            body: BedrockConverseRequestBody = {
-                "inferenceConfig": self._get_inference_config(),
-            }
-
-            if system_message:
-                body["system"] = cast(
-                    "list[SystemContentBlockTypeDef]",
-                    cast(object, [{"text": system_message}]),
+        with llm_call_context():
+            try:
+                self._emit_call_started_event(
+                    messages=messages,
+                    tools=tools,
+                    callbacks=callbacks,
+                    available_functions=available_functions,
+                    from_task=from_task,
+                    from_agent=from_agent,
                 )
 
-            # Add tool config if present or if messages contain tool content
-            # Bedrock requires toolConfig when messages have toolUse/toolResult
-            if tools:
-                tool_config: ToolConfigurationTypeDef = {
-                    "tools": cast(
-                        "Sequence[ToolTypeDef]",
-                        cast(object, self._format_tools_for_converse(tools)),
-                    )
+                formatted_messages, system_message = self._format_messages_for_converse(
+                    messages
+                )
+
+                body: BedrockConverseRequestBody = {
+                    "inferenceConfig": self._get_inference_config(),
                 }
-                body["toolConfig"] = tool_config
-            elif self._messages_contain_tool_content(formatted_messages):
-                # Create minimal toolConfig from tool history in messages
-                tools_from_history = self._extract_tools_from_message_history(
-                    formatted_messages
-                )
-                if tools_from_history:
-                    body["toolConfig"] = cast(
-                        "ToolConfigurationTypeDef",
-                        cast(object, {"tools": tools_from_history}),
+
+                if system_message:
+                    body["system"] = cast(
+                        "list[SystemContentBlockTypeDef]",
+                        cast(object, [{"text": system_message}]),
                     )
 
-            if self.guardrail_config:
-                guardrail_config: GuardrailConfigurationTypeDef = cast(
-                    "GuardrailConfigurationTypeDef", cast(object, self.guardrail_config)
-                )
-                body["guardrailConfig"] = guardrail_config
+                # Add tool config if present or if messages contain tool content
+                # Bedrock requires toolConfig when messages have toolUse/toolResult
+                if tools:
+                    tool_config: ToolConfigurationTypeDef = {
+                        "tools": cast(
+                            "Sequence[ToolTypeDef]",
+                            cast(object, self._format_tools_for_converse(tools)),
+                        )
+                    }
+                    body["toolConfig"] = tool_config
+                elif self._messages_contain_tool_content(formatted_messages):
+                    # Create minimal toolConfig from tool history in messages
+                    tools_from_history = self._extract_tools_from_message_history(
+                        formatted_messages
+                    )
+                    if tools_from_history:
+                        body["toolConfig"] = cast(
+                            "ToolConfigurationTypeDef",
+                            cast(object, {"tools": tools_from_history}),
+                        )
 
-            if self.additional_model_request_fields:
-                body["additionalModelRequestFields"] = (
-                    self.additional_model_request_fields
-                )
+                if self.guardrail_config:
+                    guardrail_config: GuardrailConfigurationTypeDef = cast(
+                        "GuardrailConfigurationTypeDef",
+                        cast(object, self.guardrail_config),
+                    )
+                    body["guardrailConfig"] = guardrail_config
 
-            if self.additional_model_response_field_paths:
-                body["additionalModelResponseFieldPaths"] = (
-                    self.additional_model_response_field_paths
-                )
+                if self.additional_model_request_fields:
+                    body["additionalModelRequestFields"] = (
+                        self.additional_model_request_fields
+                    )
 
-            if self.stream:
-                return await self._ahandle_streaming_converse(
+                if self.additional_model_response_field_paths:
+                    body["additionalModelResponseFieldPaths"] = (
+                        self.additional_model_response_field_paths
+                    )
+
+                if self.stream:
+                    return await self._ahandle_streaming_converse(
+                        formatted_messages,
+                        body,
+                        available_functions,
+                        from_task,
+                        from_agent,
+                        effective_response_model,
+                    )
+
+                return await self._ahandle_converse(
                     formatted_messages,
                     body,
                     available_functions,
@@ -585,26 +600,17 @@ class BedrockCompletion(BaseLLM):
                     effective_response_model,
                 )
 
-            return await self._ahandle_converse(
-                formatted_messages,
-                body,
-                available_functions,
-                from_task,
-                from_agent,
-                effective_response_model,
-            )
+            except Exception as e:
+                if is_context_length_exceeded(e):
+                    logging.error(f"Context window exceeded: {e}")
+                    raise LLMContextLengthExceededError(str(e)) from e
 
-        except Exception as e:
-            if is_context_length_exceeded(e):
-                logging.error(f"Context window exceeded: {e}")
-                raise LLMContextLengthExceededError(str(e)) from e
-
-            error_msg = f"AWS Bedrock API call failed: {e!s}"
-            logging.error(error_msg)
-            self._emit_call_failed_event(
-                error=error_msg, from_task=from_task, from_agent=from_agent
-            )
-            raise
+                error_msg = f"AWS Bedrock API call failed: {e!s}"
+                logging.error(error_msg)
+                self._emit_call_failed_event(
+                    error=error_msg, from_task=from_task, from_agent=from_agent
+                )
+                raise
 
     def _handle_converse(
         self,

lib/crewai/src/crewai/llms/providers/gemini/completion.py

@@ -10,7 +10,7 @@ from typing import TYPE_CHECKING, Any, Literal, cast
 from pydantic import BaseModel
 
 from crewai.events.types.llm_events import LLMCallType
-from crewai.llms.base_llm import BaseLLM
+from crewai.llms.base_llm import BaseLLM, llm_call_context
 from crewai.utilities.agent_utils import is_context_length_exceeded
 from crewai.utilities.exceptions.context_window_exceeding_exception import (
     LLMContextLengthExceededError,
@@ -293,33 +293,45 @@ class GeminiCompletion(BaseLLM):
         Returns:
             Chat completion response or tool call result
         """
-        try:
-            self._emit_call_started_event(
-                messages=messages,
-                tools=tools,
-                callbacks=callbacks,
-                available_functions=available_functions,
-                from_task=from_task,
-                from_agent=from_agent,
-            )
-            self.tools = tools
-            effective_response_model = response_model or self.response_format
+        with llm_call_context():
+            try:
+                self._emit_call_started_event(
+                    messages=messages,
+                    tools=tools,
+                    callbacks=callbacks,
+                    available_functions=available_functions,
+                    from_task=from_task,
+                    from_agent=from_agent,
+                )
+                self.tools = tools
+                effective_response_model = response_model or self.response_format
 
-            formatted_content, system_instruction = self._format_messages_for_gemini(
-                messages
-            )
+                formatted_content, system_instruction = (
+                    self._format_messages_for_gemini(messages)
+                )
 
-            messages_for_hooks = self._convert_contents_to_dict(formatted_content)
+                messages_for_hooks = self._convert_contents_to_dict(formatted_content)
 
-            if not self._invoke_before_llm_call_hooks(messages_for_hooks, from_agent):
-                raise ValueError("LLM call blocked by before_llm_call hook")
+                if not self._invoke_before_llm_call_hooks(
+                    messages_for_hooks, from_agent
+                ):
+                    raise ValueError("LLM call blocked by before_llm_call hook")
 
-            config = self._prepare_generation_config(
-                system_instruction, tools, effective_response_model
-            )
+                config = self._prepare_generation_config(
+                    system_instruction, tools, effective_response_model
+                )
 
-            if self.stream:
-                return self._handle_streaming_completion(
+                if self.stream:
+                    return self._handle_streaming_completion(
+                        formatted_content,
+                        config,
+                        available_functions,
+                        from_task,
+                        from_agent,
+                        effective_response_model,
+                    )
+
+                return self._handle_completion(
                     formatted_content,
                     config,
                     available_functions,
@@ -328,29 +340,20 @@ class GeminiCompletion(BaseLLM):
                     effective_response_model,
                 )
 
-            return self._handle_completion(
-                formatted_content,
-                config,
-                available_functions,
-                from_task,
-                from_agent,
-                effective_response_model,
-            )
-
-        except APIError as e:
-            error_msg = f"Google Gemini API error: {e.code} - {e.message}"
-            logging.error(error_msg)
-            self._emit_call_failed_event(
-                error=error_msg, from_task=from_task, from_agent=from_agent
-            )
-            raise
-        except Exception as e:
-            error_msg = f"Google Gemini API call failed: {e!s}"
-            logging.error(error_msg)
-            self._emit_call_failed_event(
-                error=error_msg, from_task=from_task, from_agent=from_agent
-            )
-            raise
+            except APIError as e:
+                error_msg = f"Google Gemini API error: {e.code} - {e.message}"
+                logging.error(error_msg)
+                self._emit_call_failed_event(
+                    error=error_msg, from_task=from_task, from_agent=from_agent
+                )
+                raise
+            except Exception as e:
+                error_msg = f"Google Gemini API call failed: {e!s}"
+                logging.error(error_msg)
+                self._emit_call_failed_event(
+                    error=error_msg, from_task=from_task, from_agent=from_agent
+                )
+                raise
 
     async def acall(
         self,
@@ -376,28 +379,38 @@ class GeminiCompletion(BaseLLM):
         Returns:
             Chat completion response or tool call result
         """
-        try:
-            self._emit_call_started_event(
-                messages=messages,
-                tools=tools,
-                callbacks=callbacks,
-                available_functions=available_functions,
-                from_task=from_task,
-                from_agent=from_agent,
-            )
-            self.tools = tools
-            effective_response_model = response_model or self.response_format
+        with llm_call_context():
+            try:
+                self._emit_call_started_event(
+                    messages=messages,
+                    tools=tools,
+                    callbacks=callbacks,
+                    available_functions=available_functions,
+                    from_task=from_task,
+                    from_agent=from_agent,
+                )
+                self.tools = tools
+                effective_response_model = response_model or self.response_format
 
-            formatted_content, system_instruction = self._format_messages_for_gemini(
-                messages
-            )
+                formatted_content, system_instruction = (
+                    self._format_messages_for_gemini(messages)
+                )
 
-            config = self._prepare_generation_config(
-                system_instruction, tools, effective_response_model
-            )
+                config = self._prepare_generation_config(
+                    system_instruction, tools, effective_response_model
+                )
 
-            if self.stream:
-                return await self._ahandle_streaming_completion(
+                if self.stream:
+                    return await self._ahandle_streaming_completion(
+                        formatted_content,
+                        config,
+                        available_functions,
+                        from_task,
+                        from_agent,
+                        effective_response_model,
+                    )
+
+                return await self._ahandle_completion(
                     formatted_content,
                     config,
                     available_functions,
@@ -406,29 +419,20 @@ class GeminiCompletion(BaseLLM):
                     effective_response_model,
                 )
 
-            return await self._ahandle_completion(
-                formatted_content,
-                config,
-                available_functions,
-                from_task,
-                from_agent,
-                effective_response_model,
-            )
-
-        except APIError as e:
-            error_msg = f"Google Gemini API error: {e.code} - {e.message}"
-            logging.error(error_msg)
-            self._emit_call_failed_event(
-                error=error_msg, from_task=from_task, from_agent=from_agent
-            )
-            raise
-        except Exception as e:
-            error_msg = f"Google Gemini API call failed: {e!s}"
-            logging.error(error_msg)
-            self._emit_call_failed_event(
-                error=error_msg, from_task=from_task, from_agent=from_agent
-            )
-            raise
+            except APIError as e:
+                error_msg = f"Google Gemini API error: {e.code} - {e.message}"
+                logging.error(error_msg)
+                self._emit_call_failed_event(
+                    error=error_msg, from_task=from_task, from_agent=from_agent
+                )
+                raise
+            except Exception as e:
+                error_msg = f"Google Gemini API call failed: {e!s}"
+                logging.error(error_msg)
+                self._emit_call_failed_event(
+                    error=error_msg, from_task=from_task, from_agent=from_agent
+                )
+                raise
 
     def _prepare_generation_config(
         self,

lib/crewai/src/crewai/llms/providers/openai/completion.py

@@ -17,7 +17,7 @@ from openai.types.responses import Response
 from pydantic import BaseModel
 
 from crewai.events.types.llm_events import LLMCallType
-from crewai.llms.base_llm import BaseLLM
+from crewai.llms.base_llm import BaseLLM, llm_call_context
 from crewai.llms.hooks.transport import AsyncHTTPTransport, HTTPTransport
 from crewai.utilities.agent_utils import is_context_length_exceeded
 from crewai.utilities.exceptions.context_window_exceeding_exception import (
@@ -382,23 +382,35 @@ class OpenAICompletion(BaseLLM):
         Returns:
             Completion response or tool call result.
         """
-        try:
-            self._emit_call_started_event(
-                messages=messages,
-                tools=tools,
-                callbacks=callbacks,
-                available_functions=available_functions,
-                from_task=from_task,
-                from_agent=from_agent,
-            )
+        with llm_call_context():
+            try:
+                self._emit_call_started_event(
+                    messages=messages,
+                    tools=tools,
+                    callbacks=callbacks,
+                    available_functions=available_functions,
+                    from_task=from_task,
+                    from_agent=from_agent,
+                )
 
-            formatted_messages = self._format_messages(messages)
+                formatted_messages = self._format_messages(messages)
 
-            if not self._invoke_before_llm_call_hooks(formatted_messages, from_agent):
-                raise ValueError("LLM call blocked by before_llm_call hook")
+                if not self._invoke_before_llm_call_hooks(
+                    formatted_messages, from_agent
+                ):
+                    raise ValueError("LLM call blocked by before_llm_call hook")
 
-            if self.api == "responses":
-                return self._call_responses(
+                if self.api == "responses":
+                    return self._call_responses(
+                        messages=formatted_messages,
+                        tools=tools,
+                        available_functions=available_functions,
+                        from_task=from_task,
+                        from_agent=from_agent,
+                        response_model=response_model,
+                    )
+
+                return self._call_completions(
                     messages=formatted_messages,
                     tools=tools,
                     available_functions=available_functions,
@@ -407,22 +419,13 @@ class OpenAICompletion(BaseLLM):
                     response_model=response_model,
                 )
 
-            return self._call_completions(
-                messages=formatted_messages,
-                tools=tools,
-                available_functions=available_functions,
-                from_task=from_task,
-                from_agent=from_agent,
-                response_model=response_model,
-            )
-
-        except Exception as e:
-            error_msg = f"OpenAI API call failed: {e!s}"
-            logging.error(error_msg)
-            self._emit_call_failed_event(
-                error=error_msg, from_task=from_task, from_agent=from_agent
-            )
-            raise
+            except Exception as e:
+                error_msg = f"OpenAI API call failed: {e!s}"
+                logging.error(error_msg)
+                self._emit_call_failed_event(
+                    error=error_msg, from_task=from_task, from_agent=from_agent
+                )
+                raise
 
     def _call_completions(
         self,
@@ -479,20 +482,30 @@ class OpenAICompletion(BaseLLM):
         Returns:
             Completion response or tool call result.
         """
-        try:
-            self._emit_call_started_event(
-                messages=messages,
-                tools=tools,
-                callbacks=callbacks,
-                available_functions=available_functions,
-                from_task=from_task,
-                from_agent=from_agent,
-            )
+        with llm_call_context():
+            try:
+                self._emit_call_started_event(
+                    messages=messages,
+                    tools=tools,
+                    callbacks=callbacks,
+                    available_functions=available_functions,
+                    from_task=from_task,
+                    from_agent=from_agent,
+                )
 
-            formatted_messages = self._format_messages(messages)
+                formatted_messages = self._format_messages(messages)
 
-            if self.api == "responses":
-                return await self._acall_responses(
+                if self.api == "responses":
+                    return await self._acall_responses(
+                        messages=formatted_messages,
+                        tools=tools,
+                        available_functions=available_functions,
+                        from_task=from_task,
+                        from_agent=from_agent,
+                        response_model=response_model,
+                    )
+
+                return await self._acall_completions(
                     messages=formatted_messages,
                     tools=tools,
                     available_functions=available_functions,
@@ -501,22 +514,13 @@ class OpenAICompletion(BaseLLM):
                     response_model=response_model,
                 )
 
-            return await self._acall_completions(
-                messages=formatted_messages,
-                tools=tools,
-                available_functions=available_functions,
-                from_task=from_task,
-                from_agent=from_agent,
-                response_model=response_model,
-            )
-
-        except Exception as e:
-            error_msg = f"OpenAI API call failed: {e!s}"
-            logging.error(error_msg)
-            self._emit_call_failed_event(
-                error=error_msg, from_task=from_task, from_agent=from_agent
-            )
-            raise
+            except Exception as e:
+                error_msg = f"OpenAI API call failed: {e!s}"
+                logging.error(error_msg)
+                self._emit_call_failed_event(
+                    error=error_msg, from_task=from_task, from_agent=from_agent
+                )
+                raise
 
     async def _acall_completions(
         self,
@@ -1521,13 +1525,16 @@ class OpenAICompletion(BaseLLM):
     ) -> list[dict[str, Any]]:
         """Convert CrewAI tool format to OpenAI function calling format."""
         from crewai.llms.providers.utils.common import safe_tool_conversion
+        from crewai.utilities.pydantic_schema_utils import (
+            force_additional_properties_false,
+        )
 
         openai_tools = []
 
         for tool in tools:
             name, description, parameters = safe_tool_conversion(tool, "OpenAI")
 
-            openai_tool = {
+            openai_tool: dict[str, Any] = {
                 "type": "function",
                 "function": {
                     "name": name,
@@ -1537,10 +1544,11 @@ class OpenAICompletion(BaseLLM):
             }
 
             if parameters:
-                if isinstance(parameters, dict):
-                    openai_tool["function"]["parameters"] = parameters  # type: ignore
-                else:
-                    openai_tool["function"]["parameters"] = dict(parameters)
+                params_dict = (
+                    parameters if isinstance(parameters, dict) else dict(parameters)
+                )
+                params_dict = force_additional_properties_false(params_dict)
+                openai_tool["function"]["parameters"] = params_dict
 
             openai_tools.append(openai_tool)
         return openai_tools

lib/crewai/src/crewai/project/crew_base.py

@@ -27,6 +27,8 @@ if TYPE_CHECKING:
     from crewai import Agent, Task
     from crewai.agents.cache.cache_handler import CacheHandler
     from crewai.crews.crew_output import CrewOutput
+    from crewai.hooks.llm_hooks import LLMCallHookContext
+    from crewai.hooks.tool_hooks import ToolCallHookContext
     from crewai.project.wrappers import (
         CrewInstance,
         OutputJsonClass,
@@ -34,6 +36,8 @@ if TYPE_CHECKING:
     )
     from crewai.tasks.task_output import TaskOutput
 
+_post_initialize_crew_hooks: list[Callable[[Any], None]] = []
+
 
 class AgentConfig(TypedDict, total=False):
     """Type definition for agent configuration dictionary.
@@ -266,6 +270,9 @@ class CrewBaseMeta(type):
         instance.map_all_agent_variables()
         instance.map_all_task_variables()
 
+        for hook in _post_initialize_crew_hooks:
+            hook(instance)
+
         original_methods = {
             name: method
             for name, method in cls.__dict__.items()
@@ -485,47 +492,61 @@ def _register_crew_hooks(instance: CrewInstance, cls: type) -> None:
             if has_agent_filter:
                 agents_filter = hook_method._filter_agents
 
-                def make_filtered_before_llm(bound_fn, agents_list):
-                    def filtered(context):
+                def make_filtered_before_llm(
+                    bound_fn: Callable[[LLMCallHookContext], bool | None],
+                    agents_list: list[str],
+                ) -> Callable[[LLMCallHookContext], bool | None]:
+                    def filtered(context: LLMCallHookContext) -> bool | None:
                         if context.agent and context.agent.role not in agents_list:
                             return None
                         return bound_fn(context)
 
                     return filtered
 
-                final_hook = make_filtered_before_llm(bound_hook, agents_filter)
+                before_llm_hook = make_filtered_before_llm(bound_hook, agents_filter)
             else:
-                final_hook = bound_hook
+                before_llm_hook = bound_hook
 
-            register_before_llm_call_hook(final_hook)
-            instance._registered_hook_functions.append(("before_llm_call", final_hook))
+            register_before_llm_call_hook(before_llm_hook)
+            instance._registered_hook_functions.append(
+                ("before_llm_call", before_llm_hook)
+            )
 
         if hasattr(hook_method, "is_after_llm_call_hook"):
             if has_agent_filter:
                 agents_filter = hook_method._filter_agents
 
-                def make_filtered_after_llm(bound_fn, agents_list):
-                    def filtered(context):
+                def make_filtered_after_llm(
+                    bound_fn: Callable[[LLMCallHookContext], str | None],
+                    agents_list: list[str],
+                ) -> Callable[[LLMCallHookContext], str | None]:
+                    def filtered(context: LLMCallHookContext) -> str | None:
                         if context.agent and context.agent.role not in agents_list:
                             return None
                         return bound_fn(context)
 
                     return filtered
 
-                final_hook = make_filtered_after_llm(bound_hook, agents_filter)
+                after_llm_hook = make_filtered_after_llm(bound_hook, agents_filter)
             else:
-                final_hook = bound_hook
+                after_llm_hook = bound_hook
 
-            register_after_llm_call_hook(final_hook)
-            instance._registered_hook_functions.append(("after_llm_call", final_hook))
+            register_after_llm_call_hook(after_llm_hook)
+            instance._registered_hook_functions.append(
+                ("after_llm_call", after_llm_hook)
+            )
 
         if hasattr(hook_method, "is_before_tool_call_hook"):
             if has_tool_filter or has_agent_filter:
                 tools_filter = getattr(hook_method, "_filter_tools", None)
                 agents_filter = getattr(hook_method, "_filter_agents", None)
 
-                def make_filtered_before_tool(bound_fn, tools_list, agents_list):
-                    def filtered(context):
+                def make_filtered_before_tool(
+                    bound_fn: Callable[[ToolCallHookContext], bool | None],
+                    tools_list: list[str] | None,
+                    agents_list: list[str] | None,
+                ) -> Callable[[ToolCallHookContext], bool | None]:
+                    def filtered(context: ToolCallHookContext) -> bool | None:
                         if tools_list and context.tool_name not in tools_list:
                             return None
                         if (
@@ -538,22 +559,28 @@ def _register_crew_hooks(instance: CrewInstance, cls: type) -> None:
 
                     return filtered
 
-                final_hook = make_filtered_before_tool(
+                before_tool_hook = make_filtered_before_tool(
                     bound_hook, tools_filter, agents_filter
                 )
             else:
-                final_hook = bound_hook
+                before_tool_hook = bound_hook
 
-            register_before_tool_call_hook(final_hook)
-            instance._registered_hook_functions.append(("before_tool_call", final_hook))
+            register_before_tool_call_hook(before_tool_hook)
+            instance._registered_hook_functions.append(
+                ("before_tool_call", before_tool_hook)
+            )
 
         if hasattr(hook_method, "is_after_tool_call_hook"):
             if has_tool_filter or has_agent_filter:
                 tools_filter = getattr(hook_method, "_filter_tools", None)
                 agents_filter = getattr(hook_method, "_filter_agents", None)
 
-                def make_filtered_after_tool(bound_fn, tools_list, agents_list):
-                    def filtered(context):
+                def make_filtered_after_tool(
+                    bound_fn: Callable[[ToolCallHookContext], str | None],
+                    tools_list: list[str] | None,
+                    agents_list: list[str] | None,
+                ) -> Callable[[ToolCallHookContext], str | None]:
+                    def filtered(context: ToolCallHookContext) -> str | None:
                         if tools_list and context.tool_name not in tools_list:
                             return None
                         if (
@@ -566,14 +593,16 @@ def _register_crew_hooks(instance: CrewInstance, cls: type) -> None:
 
                     return filtered
 
-                final_hook = make_filtered_after_tool(
+                after_tool_hook = make_filtered_after_tool(
                     bound_hook, tools_filter, agents_filter
                 )
             else:
-                final_hook = bound_hook
+                after_tool_hook = bound_hook
 
-            register_after_tool_call_hook(final_hook)
-            instance._registered_hook_functions.append(("after_tool_call", final_hook))
+            register_after_tool_call_hook(after_tool_hook)
+            instance._registered_hook_functions.append(
+                ("after_tool_call", after_tool_hook)
+            )
 
     instance._hooks_being_registered = False
 

lib/crewai/src/crewai/project/wrappers.py

@@ -72,6 +72,8 @@ class CrewInstance(Protocol):
     __crew_metadata__: CrewMetadata
     _mcp_server_adapter: Any
     _all_methods: dict[str, Callable[..., Any]]
+    _registered_hook_functions: list[tuple[str, Callable[..., Any]]]
+    _hooks_being_registered: bool
     agents: list[Agent]
     tasks: list[Task]
     base_directory: Path

lib/crewai/src/crewai/task.py

@@ -31,6 +31,8 @@ from pydantic_core import PydanticCustomError
 from typing_extensions import Self
 
 from crewai.agents.agent_builder.base_agent import BaseAgent
+from crewai.context import reset_current_task_id, set_current_task_id
+from crewai.core.providers.content_processor import process_content
 from crewai.events.event_bus import crewai_event_bus
 from crewai.events.types.task_events import (
     TaskCompletedEvent,
@@ -496,6 +498,7 @@ class Task(BaseModel):
         tools: list[BaseTool] | None = None,
     ) -> TaskOutput:
         """Execute the task synchronously."""
+        self.start_time = datetime.datetime.now()
         return self._execute_core(agent, context, tools)
 
     @property
@@ -536,6 +539,7 @@ class Task(BaseModel):
     ) -> None:
         """Execute the task asynchronously with context handling."""
         try:
+            self.start_time = datetime.datetime.now()
             result = self._execute_core(agent, context, tools)
             future.set_result(result)
         except Exception as e:
@@ -548,6 +552,7 @@ class Task(BaseModel):
         tools: list[BaseTool] | None = None,
     ) -> TaskOutput:
         """Execute the task asynchronously using native async/await."""
+        self.start_time = datetime.datetime.now()
         return await self._aexecute_core(agent, context, tools)
 
     async def _aexecute_core(
@@ -557,6 +562,7 @@ class Task(BaseModel):
         tools: list[Any] | None,
     ) -> TaskOutput:
         """Run the core execution logic of the task asynchronously."""
+        task_id_token = set_current_task_id(str(self.id))
         self._store_input_files()
         try:
             agent = agent or self.agent
@@ -566,8 +572,6 @@ class Task(BaseModel):
                     f"The task '{self.description}' has no agent assigned, therefore it can't be executed directly and should be executed in a Crew using a specific process that support that, like hierarchical."
                 )
 
-            self.start_time = datetime.datetime.now()
-
             self.prompt_context = context
             tools = tools or self.tools or []
 
@@ -579,6 +583,8 @@ class Task(BaseModel):
                 tools=tools,
             )
 
+            self._post_agent_execution(agent)
+
             if not self._guardrails and not self._guardrail:
                 pydantic_output, json_output = self._export_output(result)
             else:
@@ -644,6 +650,7 @@ class Task(BaseModel):
             raise e  # Re-raise the exception after emitting the event
         finally:
             clear_task_files(self.id)
+            reset_current_task_id(task_id_token)
 
     def _execute_core(
         self,
@@ -652,6 +659,7 @@ class Task(BaseModel):
         tools: list[Any] | None,
     ) -> TaskOutput:
         """Run the core execution logic of the task."""
+        task_id_token = set_current_task_id(str(self.id))
         self._store_input_files()
         try:
             agent = agent or self.agent
@@ -661,8 +669,6 @@ class Task(BaseModel):
                     f"The task '{self.description}' has no agent assigned, therefore it can't be executed directly and should be executed in a Crew using a specific process that support that, like hierarchical."
                 )
 
-            self.start_time = datetime.datetime.now()
-
             self.prompt_context = context
             tools = tools or self.tools or []
 
@@ -674,6 +680,8 @@ class Task(BaseModel):
                 tools=tools,
             )
 
+            self._post_agent_execution(agent)
+
             if not self._guardrails and not self._guardrail:
                 pydantic_output, json_output = self._export_output(result)
             else:
@@ -740,6 +748,10 @@ class Task(BaseModel):
             raise e  # Re-raise the exception after emitting the event
         finally:
             clear_task_files(self.id)
+            reset_current_task_id(task_id_token)
+
+    def _post_agent_execution(self, agent: BaseAgent) -> None:
+        pass
 
     def prompt(self) -> str:
         """Generates the task prompt with optional markdown formatting.
@@ -863,6 +875,11 @@ Follow these guidelines:
         except ValueError as e:
             raise ValueError(f"Error interpolating description: {e!s}") from e
 
+        self.description = process_content(self.description, {"task": self})
+        self._original_expected_output = process_content(
+            self._original_expected_output, {"task": self}
+        )
+
         try:
             self.expected_output = interpolate_only(
                 input_string=self._original_expected_output, inputs=inputs

lib/crewai/src/crewai/tasks/hallucination_guardrail.py

@@ -6,6 +6,7 @@ Classes:
     HallucinationGuardrail: Placeholder guardrail that validates task outputs.
 """
 
+from collections.abc import Callable
 from typing import Any
 
 from crewai.llm import LLM
@@ -13,32 +14,36 @@ from crewai.tasks.task_output import TaskOutput
 from crewai.utilities.logger import Logger
 
 
+_validate_output_hook: Callable[..., tuple[bool, Any]] | None = None
+
+
 class HallucinationGuardrail:
     """Placeholder for the HallucinationGuardrail feature.
 
     Attributes:
-        context: The reference context that outputs would be checked against.
+        context: Optional reference context that outputs would be checked against.
         llm: The language model that would be used for evaluation.
         threshold: Optional minimum faithfulness score that would be required to pass.
         tool_response: Optional tool response information that would be used in evaluation.
 
     Examples:
-        >>> # Basic usage with default verdict logic
+        >>> # Basic usage without context (uses task expected_output as context)
+        >>> guardrail = HallucinationGuardrail(llm=agent.llm)
+
+        >>> # With context for reference
         >>> guardrail = HallucinationGuardrail(
-        ...     context="AI helps with various tasks including analysis and generation.",
         ...     llm=agent.llm,
+        ...     context="AI helps with various tasks including analysis and generation.",
         ... )
 
         >>> # With custom threshold for stricter validation
         >>> strict_guardrail = HallucinationGuardrail(
-        ...     context="Quantum computing uses qubits in superposition.",
         ...     llm=agent.llm,
-        ...     threshold=8.0,  # Would require score >= 8 to pass in enterprise version
+        ...     threshold=8.0,  # Require score >= 8 to pass
         ... )
 
         >>> # With tool response for additional context
         >>> guardrail_with_tools = HallucinationGuardrail(
-        ...     context="The current weather data",
         ...     llm=agent.llm,
         ...     tool_response="Weather API returned: Temperature 22°C, Humidity 65%",
         ... )
@@ -46,16 +51,17 @@ class HallucinationGuardrail:
 
     def __init__(
         self,
-        context: str,
         llm: LLM,
+        context: str | None = None,
         threshold: float | None = None,
         tool_response: str = "",
     ):
         """Initialize the HallucinationGuardrail placeholder.
 
         Args:
-            context: The reference context that outputs would be checked against.
             llm: The language model that would be used for evaluation.
+            context: Optional reference context that outputs would be checked against.
+                If not provided, the task's expected_output will be used as context.
             threshold: Optional minimum faithfulness score that would be required to pass.
             tool_response: Optional tool response information that would be used in evaluation.
         """
@@ -78,16 +84,17 @@ class HallucinationGuardrail:
     def __call__(self, task_output: TaskOutput) -> tuple[bool, Any]:
         """Validate a task output against hallucination criteria.
 
-        In the open source, this method always returns that the output is valid.
-
         Args:
             task_output: The output to be validated.
 
         Returns:
             A tuple containing:
-                - True
-                - The raw task output
+                - True if validation passed, False otherwise
+                - The raw task output if valid, or error feedback if invalid
         """
+        if callable(_validate_output_hook):
+            return _validate_output_hook(self, task_output)
+
         self._logger.log(
             "warning",
             "Premium hallucination detection skipped (use for free at https://app.crewai.com)\n",

lib/crewai/src/crewai/tasks/llm_guardrail.py

@@ -1,6 +1,10 @@
+import asyncio
+from collections.abc import Coroutine
+import inspect
 from typing import Any
 
 from pydantic import BaseModel, Field
+from typing_extensions import TypeIs
 
 from crewai.agent import Agent
 from crewai.lite_agent_output import LiteAgentOutput
@@ -8,6 +12,13 @@ from crewai.llms.base_llm import BaseLLM
 from crewai.tasks.task_output import TaskOutput
 
 
+def _is_coroutine(
+    obj: LiteAgentOutput | Coroutine[Any, Any, LiteAgentOutput],
+) -> TypeIs[Coroutine[Any, Any, LiteAgentOutput]]:
+    """Check if obj is a coroutine for type narrowing."""
+    return inspect.iscoroutine(obj)
+
+
 class LLMGuardrailResult(BaseModel):
     valid: bool = Field(
         description="Whether the task output complies with the guardrail"
@@ -62,7 +73,10 @@ class LLMGuardrail:
         - If the Task result complies with the guardrail, saying that is valid
         """
 
-        return agent.kickoff(query, response_format=LLMGuardrailResult)
+        kickoff_result = agent.kickoff(query, response_format=LLMGuardrailResult)
+        if _is_coroutine(kickoff_result):
+            return asyncio.run(kickoff_result)
+        return kickoff_result
 
     def __call__(self, task_output: TaskOutput) -> tuple[bool, Any]:
         """Validates the output of a task based on specified criteria.

lib/crewai/src/crewai/telemetry/telemetry.py

@@ -903,7 +903,7 @@ class Telemetry:
                         {
                             "id": str(task.id),
                             "description": task.description,
-                            "output": task.output.raw_output,
+                            "output": task.output.raw if task.output else "",
                         }
                         for task in crew.tasks
                     ]
@@ -923,6 +923,9 @@ class Telemetry:
             value: The attribute value.
         """
 
+        if span is None:
+            return
+
         def _operation() -> None:
             return span.set_attribute(key, value)
 

lib/crewai/src/crewai/tools/tool_usage.py

@@ -270,6 +270,7 @@ class ToolUsage:
         result = None  # type: ignore
         should_retry = False
         available_tool = None
+        error_event_emitted = False
 
         try:
             if self.tools_handler and self.tools_handler.cache:
@@ -408,6 +409,7 @@ class ToolUsage:
 
                 except Exception as e:
                     self.on_tool_error(tool=tool, tool_calling=calling, e=e)
+                    error_event_emitted = True
                     self._run_attempts += 1
                     if self._run_attempts > self._max_parsing_attempts:
                         self._telemetry.tool_usage_error(llm=self.function_calling_llm)
@@ -435,7 +437,7 @@ class ToolUsage:
                 result = self._format_result(result=result)
 
         finally:
-            if started_event_emitted:
+            if started_event_emitted and not error_event_emitted:
                 self.on_tool_use_finished(
                     tool=tool,
                     tool_calling=calling,
@@ -500,6 +502,7 @@ class ToolUsage:
         result = None  # type: ignore
         should_retry = False
         available_tool = None
+        error_event_emitted = False
 
         try:
             if self.tools_handler and self.tools_handler.cache:
@@ -638,6 +641,7 @@ class ToolUsage:
 
                 except Exception as e:
                     self.on_tool_error(tool=tool, tool_calling=calling, e=e)
+                    error_event_emitted = True
                     self._run_attempts += 1
                     if self._run_attempts > self._max_parsing_attempts:
                         self._telemetry.tool_usage_error(llm=self.function_calling_llm)
@@ -665,7 +669,7 @@ class ToolUsage:
                 result = self._format_result(result=result)
 
         finally:
-            if started_event_emitted:
+            if started_event_emitted and not error_event_emitted:
                 self.on_tool_use_finished(
                     tool=tool,
                     tool_calling=calling,

lib/crewai/src/crewai/types/hitl.py

@@ -1,37 +0,0 @@
-"""Human-in-the-loop (HITL) type definitions.
-
-This module provides type definitions for human-in-the-loop interactions
-in crew executions.
-"""
-
-from typing import TypedDict
-
-
-class HITLResumeInfo(TypedDict, total=False):
-    """HITL resume information passed from flow to crew.
-
-    Attributes:
-        task_id: Unique identifier for the task.
-        crew_execution_id: Unique identifier for the crew execution.
-        task_key: Key identifying the specific task.
-        task_output: Output from the task before human intervention.
-        human_feedback: Feedback provided by the human.
-        previous_messages: History of messages in the conversation.
-    """
-
-    task_id: str
-    crew_execution_id: str
-    task_key: str
-    task_output: str
-    human_feedback: str
-    previous_messages: list[dict[str, str]]
-
-
-class CrewInputsWithHITL(TypedDict, total=False):
-    """Crew inputs that may contain HITL resume information.
-
-    Attributes:
-        _hitl_resume: Optional HITL resume information for continuing execution.
-    """
-
-    _hitl_resume: HITLResumeInfo

lib/crewai/src/crewai/utilities/agent_utils.py

@@ -42,6 +42,8 @@ if TYPE_CHECKING:
     from crewai.llm import LLM
     from crewai.task import Task
 
+_create_plus_client_hook: Callable[[], Any] | None = None
+
 
 class SummaryContent(TypedDict):
     """Structure for summary content entries.
@@ -91,7 +93,11 @@ def parse_tools(tools: list[BaseTool]) -> list[CrewStructuredTool]:
 
     for tool in tools:
         if isinstance(tool, CrewAITool):
-            tools_list.append(tool.to_structured_tool())
+            structured_tool = tool.to_structured_tool()
+            structured_tool.current_usage_count = 0
+            if structured_tool._original_tool:
+                structured_tool._original_tool.current_usage_count = 0
+            tools_list.append(structured_tool)
         else:
             raise ValueError("Tool is not a CrewStructuredTool or BaseTool")
 
@@ -818,12 +824,15 @@ def load_agent_from_repository(from_repository: str) -> dict[str, Any]:
     if from_repository:
         import importlib
 
-        from crewai.cli.authentication.token import get_auth_token
-        from crewai.cli.plus_api import PlusAPI
+        if callable(_create_plus_client_hook):
+            client = _create_plus_client_hook()
+        else:
+            from crewai.cli.authentication.token import get_auth_token
+            from crewai.cli.plus_api import PlusAPI
 
-        client = PlusAPI(api_key=get_auth_token())
+            client = PlusAPI(api_key=get_auth_token())
         _print_current_organization()
-        response = client.get_agent(from_repository)
+        response = asyncio.run(client.get_agent(from_repository))
         if response.status_code == 404:
             raise AgentRepositoryError(
                 f"Agent {from_repository} does not exist, make sure the name is correct or the agent is available on your organization."

lib/crewai/src/crewai/utilities/evaluators/crew_evaluator_handler.py

@@ -1,7 +1,7 @@
 from __future__ import annotations
 
 from collections import defaultdict
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, Any
 
 from pydantic import BaseModel, Field, InstanceOf
 from rich.box import HEAVY_EDGE
@@ -36,7 +36,13 @@ class CrewEvaluator:
         iteration: The current iteration of the evaluation.
     """
 
-    def __init__(self, crew: Crew, eval_llm: InstanceOf[BaseLLM]) -> None:
+    def __init__(
+        self,
+        crew: Crew,
+        eval_llm: InstanceOf[BaseLLM] | str | None = None,
+        openai_model_name: str | None = None,
+        llm: InstanceOf[BaseLLM] | str | None = None,
+    ) -> None:
         self.crew = crew
         self.llm = eval_llm
         self.tasks_scores: defaultdict[int, list[float]] = defaultdict(list)
@@ -86,7 +92,9 @@ class CrewEvaluator:
         """
         self.iteration = iteration
 
-    def print_crew_evaluation_result(self) -> None:
+    def print_crew_evaluation_result(
+        self, token_usage: list[dict[str, Any]] | None = None
+    ) -> None:
         """
         Prints the evaluation result of the crew in a table.
         A Crew with 2 tasks using the command crewai test -n 3
@@ -204,7 +212,7 @@ class CrewEvaluator:
                 CrewTestResultEvent(
                     quality=quality_score,
                     execution_duration=current_task.execution_duration,
-                    model=self.llm.model,
+                    model=getattr(self.llm, "model", str(self.llm)),
                     crew_name=self.crew.name,
                     crew=self.crew,
                 ),

lib/crewai/src/crewai/utilities/printer.py

@@ -4,6 +4,8 @@ from __future__ import annotations
 
 from typing import TYPE_CHECKING, Final, Literal, NamedTuple
 
+from crewai.events.utils.console_formatter import should_suppress_console_output
+
 
 if TYPE_CHECKING:
     from _typeshed import SupportsWrite
@@ -77,6 +79,8 @@ class Printer:
             file: A file-like object (stream); defaults to the current sys.stdout.
             flush: Whether to forcibly flush the stream.
         """
+        if should_suppress_console_output():
+            return
         if isinstance(content, str):
             content = [ColoredText(content, color)]
         print(

lib/crewai/src/crewai/utilities/pydantic_schema_utils.py

@@ -19,9 +19,10 @@ from collections.abc import Callable
 from copy import deepcopy
 import datetime
 import logging
-from typing import TYPE_CHECKING, Annotated, Any, Literal, Union
+from typing import TYPE_CHECKING, Annotated, Any, Final, Literal, TypedDict, Union
 import uuid
 
+import jsonref  # type: ignore[import-untyped]
 from pydantic import (
     UUID1,
     UUID3,
@@ -69,6 +70,21 @@ else:
         EmailStr = str
 
 
+class JsonSchemaInfo(TypedDict):
+    """Inner structure for JSON schema metadata."""
+
+    name: str
+    strict: Literal[True]
+    schema: dict[str, Any]
+
+
+class ModelDescription(TypedDict):
+    """Return type for generate_model_description."""
+
+    type: Literal["json_schema"]
+    json_schema: JsonSchemaInfo
+
+
 def resolve_refs(schema: dict[str, Any]) -> dict[str, Any]:
     """Recursively resolve all local $refs in the given JSON Schema using $defs as the source.
 
@@ -127,6 +143,102 @@ def add_key_in_dict_recursively(
     return d
 
 
+def force_additional_properties_false(d: Any) -> Any:
+    """Force additionalProperties=false on all object-type dicts recursively.
+
+    OpenAI strict mode requires all objects to have additionalProperties=false.
+    This function overwrites any existing value to ensure compliance.
+
+    Also ensures objects have properties and required arrays, even if empty,
+    as OpenAI strict mode requires these for all object types.
+
+    Args:
+        d: The dictionary/list to modify.
+
+    Returns:
+        The modified dictionary/list.
+    """
+    if isinstance(d, dict):
+        if d.get("type") == "object":
+            d["additionalProperties"] = False
+            if "properties" not in d:
+                d["properties"] = {}
+            if "required" not in d:
+                d["required"] = []
+        for v in d.values():
+            force_additional_properties_false(v)
+    elif isinstance(d, list):
+        for i in d:
+            force_additional_properties_false(i)
+    return d
+
+
+OPENAI_SUPPORTED_FORMATS: Final[
+    set[Literal["date-time", "date", "time", "duration"]]
+] = {
+    "date-time",
+    "date",
+    "time",
+    "duration",
+}
+
+
+def strip_unsupported_formats(d: Any) -> Any:
+    """Remove format annotations that OpenAI strict mode doesn't support.
+
+    OpenAI only supports: date-time, date, time, duration.
+    Other formats like uri, email, uuid etc. cause validation errors.
+
+    Args:
+        d: The dictionary/list to modify.
+
+    Returns:
+        The modified dictionary/list.
+    """
+    if isinstance(d, dict):
+        format_value = d.get("format")
+        if (
+            isinstance(format_value, str)
+            and format_value not in OPENAI_SUPPORTED_FORMATS
+        ):
+            del d["format"]
+        for v in d.values():
+            strip_unsupported_formats(v)
+    elif isinstance(d, list):
+        for i in d:
+            strip_unsupported_formats(i)
+    return d
+
+
+def ensure_type_in_schemas(d: Any) -> Any:
+    """Ensure all schema objects in anyOf/oneOf have a 'type' key.
+
+    OpenAI strict mode requires every schema to have a 'type' key.
+    Empty schemas {} in anyOf/oneOf are converted to {"type": "object"}.
+
+    Args:
+        d: The dictionary/list to modify.
+
+    Returns:
+        The modified dictionary/list.
+    """
+    if isinstance(d, dict):
+        for key in ("anyOf", "oneOf"):
+            if key in d:
+                schema_list = d[key]
+                for i, schema in enumerate(schema_list):
+                    if isinstance(schema, dict) and schema == {}:
+                        schema_list[i] = {"type": "object"}
+                    else:
+                        ensure_type_in_schemas(schema)
+        for v in d.values():
+            ensure_type_in_schemas(v)
+    elif isinstance(d, list):
+        for item in d:
+            ensure_type_in_schemas(item)
+    return d
+
+
 def fix_discriminator_mappings(schema: dict[str, Any]) -> dict[str, Any]:
     """Replace '#/$defs/...' references in discriminator.mapping with just the model name.
 
@@ -263,7 +375,49 @@ def ensure_all_properties_required(schema: dict[str, Any]) -> dict[str, Any]:
     return schema
 
 
-def generate_model_description(model: type[BaseModel]) -> dict[str, Any]:
+def strip_null_from_types(schema: dict[str, Any]) -> dict[str, Any]:
+    """Remove null type from anyOf/type arrays.
+
+    Pydantic generates `T | None` for optional fields, which creates schemas with
+    null in the type. However, for MCP tools, optional fields should be omitted
+    entirely rather than sent as null. This function strips null from types.
+
+    Args:
+        schema: JSON schema dictionary.
+
+    Returns:
+        Modified schema with null types removed.
+    """
+    if isinstance(schema, dict):
+        if "anyOf" in schema:
+            any_of = schema["anyOf"]
+            non_null = [opt for opt in any_of if opt.get("type") != "null"]
+            if len(non_null) == 1:
+                schema.pop("anyOf")
+                schema.update(non_null[0])
+            elif len(non_null) > 1:
+                schema["anyOf"] = non_null
+
+        type_value = schema.get("type")
+        if isinstance(type_value, list) and "null" in type_value:
+            non_null_types = [t for t in type_value if t != "null"]
+            if len(non_null_types) == 1:
+                schema["type"] = non_null_types[0]
+            elif len(non_null_types) > 1:
+                schema["type"] = non_null_types
+
+        for value in schema.values():
+            if isinstance(value, dict):
+                strip_null_from_types(value)
+            elif isinstance(value, list):
+                for item in value:
+                    if isinstance(item, dict):
+                        strip_null_from_types(item)
+
+    return schema
+
+
+def generate_model_description(model: type[BaseModel]) -> ModelDescription:
     """Generate JSON schema description of a Pydantic model.
 
     This function takes a Pydantic model class and returns its JSON schema,
@@ -274,17 +428,13 @@ def generate_model_description(model: type[BaseModel]) -> dict[str, Any]:
         model: A Pydantic model class.
 
     Returns:
-        A JSON schema dictionary representation of the model.
+        A ModelDescription with JSON schema representation of the model.
     """
     json_schema = model.model_json_schema(ref_template="#/$defs/{model}")
 
-    json_schema = add_key_in_dict_recursively(
-        json_schema,
-        key="additionalProperties",
-        value=False,
-        criteria=lambda d: d.get("type") == "object"
-        and "additionalProperties" not in d,
-    )
+    json_schema = force_additional_properties_false(json_schema)
+    json_schema = strip_unsupported_formats(json_schema)
+    json_schema = ensure_type_in_schemas(json_schema)
 
     json_schema = resolve_refs(json_schema)
 
@@ -292,6 +442,7 @@ def generate_model_description(model: type[BaseModel]) -> dict[str, Any]:
     json_schema = fix_discriminator_mappings(json_schema)
     json_schema = convert_oneof_to_anyof(json_schema)
     json_schema = ensure_all_properties_required(json_schema)
+    json_schema = strip_null_from_types(json_schema)
 
     return {
         "type": "json_schema",
@@ -376,14 +527,19 @@ def create_model_from_schema(  # type: ignore[no-any-unimported]
         >>> person.name
         'John'
     """
+    json_schema = dict(jsonref.replace_refs(json_schema, proxies=False))
+
     effective_root = root_schema or json_schema
 
+    json_schema = force_additional_properties_false(json_schema)
+    effective_root = force_additional_properties_false(effective_root)
+
     if "allOf" in json_schema:
         json_schema = _merge_all_of_schemas(json_schema["allOf"], effective_root)
         if "title" not in json_schema and "title" in (root_schema or {}):
             json_schema["title"] = (root_schema or {}).get("title")
 
-    model_name = json_schema.get("title", "DynamicModel")
+    model_name = json_schema.get("title") or "DynamicModel"
     field_definitions = {
         name: _json_schema_to_pydantic_field(
             name, prop, json_schema.get("required", []), effective_root
@@ -391,9 +547,11 @@ def create_model_from_schema(  # type: ignore[no-any-unimported]
         for name, prop in (json_schema.get("properties", {}) or {}).items()
     }
 
+    effective_config = __config__ or ConfigDict(extra="forbid")
+
     return create_model_base(
         model_name,
-        __config__=__config__,
+        __config__=effective_config,
         __base__=__base__,
         __module__=__module__,
         __validators__=__validators__,
@@ -572,8 +730,10 @@ def _json_schema_to_pydantic_type(
         any_of_schemas = json_schema.get("anyOf", []) + json_schema.get("oneOf", [])
     if any_of_schemas:
         any_of_types = [
-            _json_schema_to_pydantic_type(schema, root_schema)
-            for schema in any_of_schemas
+            _json_schema_to_pydantic_type(
+                schema, root_schema, name_=f"{name_ or 'Union'}Option{i}"
+            )
+            for i, schema in enumerate(any_of_schemas)
         ]
         return Union[tuple(any_of_types)]  # noqa: UP007
 
@@ -609,7 +769,7 @@ def _json_schema_to_pydantic_type(
         if properties:
             json_schema_ = json_schema.copy()
             if json_schema_.get("title") is None:
-                json_schema_["title"] = name_
+                json_schema_["title"] = name_ or "DynamicModel"
             return create_model_from_schema(json_schema_, root_schema=root_schema)
         return dict
     if type_ == "null":

lib/crewai/src/crewai/utilities/serialization.py

@@ -19,6 +19,7 @@ def to_serializable(
     exclude: set[str] | None = None,
     max_depth: int = 5,
     _current_depth: int = 0,
+    _ancestors: set[int] | None = None,
 ) -> Serializable:
     """Converts a Python object into a JSON-compatible representation.
 
@@ -31,6 +32,7 @@ def to_serializable(
         exclude: Set of keys to exclude from the result.
         max_depth: Maximum recursion depth. Defaults to 5.
         _current_depth: Current recursion depth (for internal use).
+        _ancestors: Set of ancestor object ids for cycle detection (for internal use).
 
     Returns:
         Serializable: A JSON-compatible structure.
@@ -41,16 +43,29 @@ def to_serializable(
     if exclude is None:
         exclude = set()
 
+    if _ancestors is None:
+        _ancestors = set()
+
     if isinstance(obj, (str, int, float, bool, type(None))):
         return obj
     if isinstance(obj, uuid.UUID):
         return str(obj)
     if isinstance(obj, (date, datetime)):
         return obj.isoformat()
+
+    object_id = id(obj)
+    if object_id in _ancestors:
+        return f"<circular_ref:{type(obj).__name__}>"
+    new_ancestors = _ancestors | {object_id}
+
     if isinstance(obj, (list, tuple, set)):
         return [
             to_serializable(
-                item, max_depth=max_depth, _current_depth=_current_depth + 1
+                item,
+                exclude=exclude,
+                max_depth=max_depth,
+                _current_depth=_current_depth + 1,
+                _ancestors=new_ancestors,
             )
             for item in obj
         ]
@@ -61,6 +76,7 @@ def to_serializable(
                 exclude=exclude,
                 max_depth=max_depth,
                 _current_depth=_current_depth + 1,
+                _ancestors=new_ancestors,
             )
             for key, value in obj.items()
             if key not in exclude
@@ -71,12 +87,16 @@ def to_serializable(
                 obj=obj.model_dump(exclude=exclude),
                 max_depth=max_depth,
                 _current_depth=_current_depth + 1,
+                _ancestors=new_ancestors,
             )
         except Exception:
             try:
                 return {
                     _to_serializable_key(k): to_serializable(
-                        v, max_depth=max_depth, _current_depth=_current_depth + 1
+                        v,
+                        max_depth=max_depth,
+                        _current_depth=_current_depth + 1,
+                        _ancestors=new_ancestors,
                     )
                     for k, v in obj.__dict__.items()
                     if k not in (exclude or set())

lib/crewai/tests/agents/agent_adapters/test_base_agent_adapter.py

@@ -51,6 +51,10 @@ class ConcreteAgentAdapter(BaseAgentAdapter):
         # Dummy implementation for MCP tools
         return []
 
+    def configure_structured_output(self, task: Any) -> None:
+        # Dummy implementation for structured output
+        pass
+
     async def aexecute_task(
         self,
         task: Any,

lib/crewai/tests/agents/test_agent.py

@@ -703,6 +703,8 @@ def test_agent_definition_based_on_dict():
 # test for human input
 @pytest.mark.vcr()
 def test_agent_human_input():
+    from crewai.core.providers.human_input import SyncHumanInputProvider
+
     # Agent configuration
     config = {
         "role": "test role",
@@ -720,7 +722,7 @@ def test_agent_human_input():
         human_input=True,
     )
 
-    # Side effect function for _ask_human_input to simulate multiple feedback iterations
+    # Side effect function for _prompt_input to simulate multiple feedback iterations
     feedback_responses = iter(
         [
             "Don't say hi, say Hello instead!",  # First feedback: instruct change
@@ -728,16 +730,16 @@ def test_agent_human_input():
         ]
     )
 
-    def ask_human_input_side_effect(*args, **kwargs):
+    def prompt_input_side_effect(*args, **kwargs):
         return next(feedback_responses)
 
-    # Patch both _ask_human_input and _invoke_loop to avoid real API/network calls.
+    # Patch both _prompt_input on provider and _invoke_loop to avoid real API/network calls.
     with (
         patch.object(
-            CrewAgentExecutor,
-            "_ask_human_input",
-            side_effect=ask_human_input_side_effect,
-        ) as mock_human_input,
+            SyncHumanInputProvider,
+            "_prompt_input",
+            side_effect=prompt_input_side_effect,
+        ) as mock_prompt_input,
         patch.object(
             CrewAgentExecutor,
             "_invoke_loop",
@@ -749,7 +751,7 @@ def test_agent_human_input():
 
         # Assertions to ensure the agent behaves correctly.
         # It should have requested feedback twice.
-        assert mock_human_input.call_count == 2
+        assert mock_prompt_input.call_count == 2
         # The final result should be processed to "Hello"
         assert output.strip().lower() == "hello"
 

lib/crewai/tests/agents/test_async_agent_executor.py

@@ -235,8 +235,13 @@ class TestAsyncAgentExecutor:
         mock_crew: MagicMock, mock_tools_handler: MagicMock
     ) -> None:
         """Test that multiple ainvoke calls can run concurrently."""
+        max_concurrent = 0
+        current_concurrent = 0
+        lock = asyncio.Lock()
 
         async def create_and_run_executor(executor_id: int) -> dict[str, Any]:
+            nonlocal max_concurrent, current_concurrent
+
             executor = CrewAgentExecutor(
                 llm=mock_llm,
                 task=mock_task,
@@ -252,7 +257,13 @@ class TestAsyncAgentExecutor:
             )
 
             async def delayed_response(*args: Any, **kwargs: Any) -> str:
-                await asyncio.sleep(0.05)
+                nonlocal max_concurrent, current_concurrent
+                async with lock:
+                    current_concurrent += 1
+                    max_concurrent = max(max_concurrent, current_concurrent)
+                await asyncio.sleep(0.01)
+                async with lock:
+                    current_concurrent -= 1
                 return f"Thought: Done\nFinal Answer: Result from executor {executor_id}"
 
             with patch(
@@ -273,19 +284,15 @@ class TestAsyncAgentExecutor:
                                         }
                                     )
 
-        import time
-
-        start = time.time()
         results = await asyncio.gather(
             create_and_run_executor(1),
             create_and_run_executor(2),
             create_and_run_executor(3),
         )
-        elapsed = time.time() - start
 
         assert len(results) == 3
         assert all("output" in r for r in results)
-        assert elapsed < 0.15, f"Expected concurrent execution, took {elapsed}s"
+        assert max_concurrent > 1, f"Expected concurrent execution, max concurrent was {max_concurrent}"
 
 
 class TestAsyncLLMResponseHelper:

lib/crewai/tests/agents/test_lite_agent.py

@@ -299,14 +299,16 @@ class TestFlow(Flow):
         return agent.kickoff("Test query")
 
 
-def verify_agent_parent_flow(result, agent, flow):
-    """Verify that both the result and agent have the correct parent flow."""
-    assert result.parent_flow is flow
+def verify_agent_flow_context(result, agent, flow):
+    """Verify that both the result and agent have the correct flow context."""
+    assert result._flow_id == flow.flow_id  # type: ignore[attr-defined]
+    assert result._request_id == flow.flow_id  # type: ignore[attr-defined]
     assert agent is not None
-    assert agent.parent_flow is flow
+    assert agent._flow_id == flow.flow_id  # type: ignore[attr-defined]
+    assert agent._request_id == flow.flow_id  # type: ignore[attr-defined]
 
 
-def test_sets_parent_flow_when_inside_flow():
+def test_sets_flow_context_when_inside_flow():
     """Test that an Agent can be created and executed inside a Flow context."""
     captured_event = None
 
@@ -604,9 +606,10 @@ def test_lite_agent_with_invalid_llm():
 
 
 @patch.dict("os.environ", {"CREWAI_PLATFORM_INTEGRATION_TOKEN": "test_token"})
+@patch("crewai_tools.tools.crewai_platform_tools.crewai_platform_action_tool.requests.post")
 @patch("crewai_tools.tools.crewai_platform_tools.crewai_platform_tool_builder.requests.get")
 @pytest.mark.vcr()
-def test_agent_kickoff_with_platform_tools(mock_get):
+def test_agent_kickoff_with_platform_tools(mock_get, mock_post):
     """Test that Agent.kickoff() properly integrates platform tools with LiteAgent"""
     mock_response = Mock()
     mock_response.raise_for_status.return_value = None
@@ -630,6 +633,15 @@ def test_agent_kickoff_with_platform_tools(mock_get):
     }
     mock_get.return_value = mock_response
 
+    # Mock the platform tool execution
+    mock_post_response = Mock()
+    mock_post_response.ok = True
+    mock_post_response.json.return_value = {
+        "success": True,
+        "issue_url": "https://github.com/test/repo/issues/1"
+    }
+    mock_post.return_value = mock_post_response
+
     agent = Agent(
         role="Test Agent",
         goal="Test goal",

lib/crewai/tests/cassettes/agents/test_agent_kickoff_with_platform_tools.yaml

@@ -1,98 +1,227 @@
 interactions:
 - request:
-    body: '{"messages": [{"role": "system", "content": "You are Test Agent. Test backstory\nYour personal goal is: Test goal\n\nYou ONLY have access to the following tools, and should NEVER make up tools that are not listed here:\n\nTool Name: create_issue\nTool Arguments: {''title'': {''description'': ''Issue title'', ''type'': ''str''}, ''body'': {''description'': ''Issue body'', ''type'': ''Union[str, NoneType]''}}\nTool Description: Create a GitHub issue\nDetailed Parameter Structure:\nObject with properties:\n  - title: Issue title (required)\n  - body: Issue body (optional)\n\nIMPORTANT: Use the following format in your response:\n\n```\nThought: you should always think about what to do\nAction: the action to take, only one name of [create_issue], just the name, exactly as it''s written.\nAction Input: the input to the action, just a simple JSON object, enclosed in curly braces, using \" to wrap keys and values.\nObservation: the result of the action\n```\n\nOnce all necessary information
-      is gathered, return the following format:\n\n```\nThought: I now know the final answer\nFinal Answer: the final answer to the original input question\n```"}, {"role": "user", "content": "Create a GitHub issue"}], "model": "gpt-3.5-turbo", "stream": false}'
+    body: '{"messages":[{"role":"system","content":"You are Test Agent. Test backstory\nYour
+      personal goal is: Test goal"},{"role":"user","content":"\nCurrent Task: Create
+      a GitHub issue"}],"model":"gpt-3.5-turbo","tool_choice":"auto","tools":[{"type":"function","function":{"name":"create_issue","description":"Create
+      a GitHub issue","strict":true,"parameters":{"additionalProperties":false,"properties":{"title":{"description":"Issue
+      title","title":"Title","type":"string"},"body":{"default":null,"description":"Issue
+      body","title":"Body","type":"string"}},"required":["title","body"],"type":"object"}}}]}'
     headers:
+      User-Agent:
+      - X-USER-AGENT-XXX
       accept:
       - application/json
       accept-encoding:
-      - gzip, deflate
+      - ACCEPT-ENCODING-XXX
+      authorization:
+      - AUTHORIZATION-XXX
       connection:
       - keep-alive
       content-length:
-      - '1233'
+      - '596'
       content-type:
       - application/json
       host:
       - api.openai.com
-      user-agent:
-      - OpenAI/Python 1.109.1
       x-stainless-arch:
-      - arm64
+      - X-STAINLESS-ARCH-XXX
       x-stainless-async:
       - 'false'
       x-stainless-lang:
       - python
       x-stainless-os:
-      - MacOS
+      - X-STAINLESS-OS-XXX
       x-stainless-package-version:
-      - 1.109.1
+      - 1.83.0
       x-stainless-read-timeout:
-      - '600'
+      - X-STAINLESS-READ-TIMEOUT-XXX
       x-stainless-retry-count:
       - '0'
       x-stainless-runtime:
       - CPython
       x-stainless-runtime-version:
-      - 3.13.3
+      - 3.13.5
     method: POST
     uri: https://api.openai.com/v1/chat/completions
   response:
     body:
-      string: "{\n  \"id\": \"chatcmpl-CULxKTEIB85AVItcEQ09z4Xi0JCID\",\n  \"object\": \"chat.completion\",\n  \"created\": 1761350274,\n  \"model\": \"gpt-3.5-turbo-0125\",\n  \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\": \"assistant\",\n        \"content\": \"I will need more specific information to create a GitHub issue. Could you please provide more details such as the title and body of the issue you would like to create?\",\n        \"refusal\": null,\n        \"annotations\": []\n      },\n      \"logprobs\": null,\n      \"finish_reason\": \"stop\"\n    }\n  ],\n  \"usage\": {\n    \"prompt_tokens\": 255,\n    \"completion_tokens\": 33,\n    \"total_tokens\": 288,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\": 0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\": {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\": 0,\n      \"rejected_prediction_tokens\": 0\n   \
-        \ }\n  },\n  \"service_tier\": \"default\",\n  \"system_fingerprint\": null\n}\n"
+      string: "{\n  \"id\": \"chatcmpl-D6L3fqygkUIZ3bN4wvSpAhdaSk7MF\",\n  \"object\":
+        \"chat.completion\",\n  \"created\": 1770403287,\n  \"model\": \"gpt-3.5-turbo-0125\",\n
+        \ \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\":
+        \"assistant\",\n        \"content\": null,\n        \"tool_calls\": [\n          {\n
+        \           \"id\": \"call_RuWuYzjzgRL3byVGhLlPi0rq\",\n            \"type\":
+        \"function\",\n            \"function\": {\n              \"name\": \"create_issue\",\n
+        \             \"arguments\": \"{\\\"title\\\":\\\"Test issue\\\",\\\"body\\\":\\\"This
+        is a test issue created for testing purposes.\\\"}\"\n            }\n          }\n
+        \       ],\n        \"refusal\": null,\n        \"annotations\": []\n      },\n
+        \     \"logprobs\": null,\n      \"finish_reason\": \"tool_calls\"\n    }\n
+        \ ],\n  \"usage\": {\n    \"prompt_tokens\": 93,\n    \"completion_tokens\":
+        28,\n    \"total_tokens\": 121,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\":
+        0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\":
+        {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\":
+        0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\":
+        \"default\",\n  \"system_fingerprint\": null\n}\n"
     headers:
       CF-RAY:
-      - 993d6b4be9862379-SJC
+      - CF-RAY-XXX
       Connection:
       - keep-alive
       Content-Type:
       - application/json
       Date:
-      - Fri, 24 Oct 2025 23:57:54 GMT
+      - Fri, 06 Feb 2026 18:41:28 GMT
       Server:
       - cloudflare
       Set-Cookie:
-      - __cf_bm=WY9bgemMDI_hUYISAPlQ2a.DBGeZfM6AjVEa3SKNg1c-1761350274-1.0.1.1-K3Qm2cl6IlDAgmocoKZ8IMUTmue6Q81hH9stECprUq_SM8LF8rR9d1sHktvRCN3.jEM.twEuFFYDNpBnN8NBRJFZcea1yvpm8Uo0G_UhyDs; path=/; expires=Sat, 25-Oct-25 00:27:54 GMT; domain=.api.openai.com; HttpOnly; Secure; SameSite=None
-      - _cfuvid=JklLS4i3hBGELpS9cz1KMpTbj72hCwP41LyXDSxWIv8-1761350274521-0.0.1.1-604800000; path=/; domain=.api.openai.com; HttpOnly; Secure; SameSite=None
+      - SET-COOKIE-XXX
       Strict-Transport-Security:
-      - max-age=31536000; includeSubDomains; preload
+      - STS-XXX
       Transfer-Encoding:
       - chunked
       X-Content-Type-Options:
-      - nosniff
+      - X-CONTENT-TYPE-XXX
       access-control-expose-headers:
-      - X-Request-ID
+      - ACCESS-CONTROL-XXX
       alt-svc:
       - h3=":443"; ma=86400
       cf-cache-status:
       - DYNAMIC
       openai-organization:
-      - crewai-iuxna1
+      - OPENAI-ORG-XXX
       openai-processing-ms:
-      - '487'
+      - '1406'
       openai-project:
-      - proj_xitITlrFeen7zjNSzML82h9x
+      - OPENAI-PROJECT-XXX
       openai-version:
       - '2020-10-01'
-      x-envoy-upstream-service-time:
-      - '526'
       x-openai-proxy-wasm:
       - v0.1
       x-ratelimit-limit-requests:
-      - '10000'
+      - X-RATELIMIT-LIMIT-REQUESTS-XXX
       x-ratelimit-limit-tokens:
-      - '50000000'
+      - X-RATELIMIT-LIMIT-TOKENS-XXX
       x-ratelimit-remaining-requests:
-      - '9999'
+      - X-RATELIMIT-REMAINING-REQUESTS-XXX
       x-ratelimit-remaining-tokens:
-      - '49999727'
+      - X-RATELIMIT-REMAINING-TOKENS-XXX
       x-ratelimit-reset-requests:
-      - 6ms
+      - X-RATELIMIT-RESET-REQUESTS-XXX
       x-ratelimit-reset-tokens:
-      - 0s
+      - X-RATELIMIT-RESET-TOKENS-XXX
       x-request-id:
-      - req_1708dc0928c64882aaa5bc2c168c140f
+      - X-REQUEST-ID-XXX
+    status:
+      code: 200
+      message: OK
+- request:
+    body: '{"messages":[{"role":"system","content":"You are Test Agent. Test backstory\nYour
+      personal goal is: Test goal"},{"role":"user","content":"\nCurrent Task: Create
+      a GitHub issue"},{"role":"assistant","content":null,"tool_calls":[{"id":"call_RuWuYzjzgRL3byVGhLlPi0rq","type":"function","function":{"name":"create_issue","arguments":"{\"title\":\"Test
+      issue\",\"body\":\"This is a test issue created for testing purposes.\"}"}}]},{"role":"tool","tool_call_id":"call_RuWuYzjzgRL3byVGhLlPi0rq","name":"create_issue","content":"{\n  \"success\":
+      true,\n  \"issue_url\": \"https://github.com/test/repo/issues/1\"\n}"}],"model":"gpt-3.5-turbo","tool_choice":"auto","tools":[{"type":"function","function":{"name":"create_issue","description":"Create
+      a GitHub issue","strict":true,"parameters":{"additionalProperties":false,"properties":{"title":{"description":"Issue
+      title","title":"Title","type":"string"},"body":{"default":null,"description":"Issue
+      body","title":"Body","type":"string"}},"required":["title","body"],"type":"object"}}}]}'
+    headers:
+      User-Agent:
+      - X-USER-AGENT-XXX
+      accept:
+      - application/json
+      accept-encoding:
+      - ACCEPT-ENCODING-XXX
+      authorization:
+      - AUTHORIZATION-XXX
+      connection:
+      - keep-alive
+      content-length:
+      - '1028'
+      content-type:
+      - application/json
+      cookie:
+      - COOKIE-XXX
+      host:
+      - api.openai.com
+      x-stainless-arch:
+      - X-STAINLESS-ARCH-XXX
+      x-stainless-async:
+      - 'false'
+      x-stainless-lang:
+      - python
+      x-stainless-os:
+      - X-STAINLESS-OS-XXX
+      x-stainless-package-version:
+      - 1.83.0
+      x-stainless-read-timeout:
+      - X-STAINLESS-READ-TIMEOUT-XXX
+      x-stainless-retry-count:
+      - '0'
+      x-stainless-runtime:
+      - CPython
+      x-stainless-runtime-version:
+      - 3.13.5
+    method: POST
+    uri: https://api.openai.com/v1/chat/completions
+  response:
+    body:
+      string: "{\n  \"id\": \"chatcmpl-D6L3hfuBxk36LIb3ekD1IVwFD5VVL\",\n  \"object\":
+        \"chat.completion\",\n  \"created\": 1770403289,\n  \"model\": \"gpt-3.5-turbo-0125\",\n
+        \ \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\":
+        \"assistant\",\n        \"content\": \"I have successfully created a GitHub
+        issue for testing purposes. You can view the issue at this URL: [Test issue](https://github.com/test/repo/issues/1)\",\n
+        \       \"refusal\": null,\n        \"annotations\": []\n      },\n      \"logprobs\":
+        null,\n      \"finish_reason\": \"stop\"\n    }\n  ],\n  \"usage\": {\n    \"prompt_tokens\":
+        154,\n    \"completion_tokens\": 36,\n    \"total_tokens\": 190,\n    \"prompt_tokens_details\":
+        {\n      \"cached_tokens\": 0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\":
+        {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\":
+        0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\":
+        \"default\",\n  \"system_fingerprint\": null\n}\n"
+    headers:
+      CF-RAY:
+      - CF-RAY-XXX
+      Connection:
+      - keep-alive
+      Content-Type:
+      - application/json
+      Date:
+      - Fri, 06 Feb 2026 18:41:29 GMT
+      Server:
+      - cloudflare
+      Strict-Transport-Security:
+      - STS-XXX
+      Transfer-Encoding:
+      - chunked
+      X-Content-Type-Options:
+      - X-CONTENT-TYPE-XXX
+      access-control-expose-headers:
+      - ACCESS-CONTROL-XXX
+      alt-svc:
+      - h3=":443"; ma=86400
+      cf-cache-status:
+      - DYNAMIC
+      openai-organization:
+      - OPENAI-ORG-XXX
+      openai-processing-ms:
+      - '888'
+      openai-project:
+      - OPENAI-PROJECT-XXX
+      openai-version:
+      - '2020-10-01'
+      x-openai-proxy-wasm:
+      - v0.1
+      x-ratelimit-limit-requests:
+      - X-RATELIMIT-LIMIT-REQUESTS-XXX
+      x-ratelimit-limit-tokens:
+      - X-RATELIMIT-LIMIT-TOKENS-XXX
+      x-ratelimit-remaining-requests:
+      - X-RATELIMIT-REMAINING-REQUESTS-XXX
+      x-ratelimit-remaining-tokens:
+      - X-RATELIMIT-REMAINING-TOKENS-XXX
+      x-ratelimit-reset-requests:
+      - X-RATELIMIT-RESET-REQUESTS-XXX
+      x-ratelimit-reset-tokens:
+      - X-RATELIMIT-RESET-TOKENS-XXX
+      x-request-id:
+      - X-REQUEST-ID-XXX
     status:
       code: 200
       message: OK

lib/crewai/tests/cassettes/agents/test_guardrail_reached_attempt_limit.yaml

@@ -1,400 +1,428 @@
 interactions:
 - request:
-    body: '{"messages": [{"role": "system", "content": "You are Sports Analyst. You are an expert at gathering and organizing information. You carefully collect details and present them in a structured way.\nYour personal goal is: Gather information about the best soccer players\n\nTo give my best complete final answer to the task respond using the exact following format:\n\nThought: I now can give a great answer\nFinal Answer: Your final answer must be the great and the most complete as possible, it must be outcome described.\n\nI MUST use these formats, my job depends on it!"}, {"role": "user", "content": "Top 10 best players in the world?"}], "model": "gpt-4o-mini", "stop": ["\nObservation:"]}'
+    body: '{"messages":[{"role":"system","content":"You are Sports Analyst. You are
+      an expert at gathering and organizing information. You carefully collect details
+      and present them in a structured way.\nYour personal goal is: Gather information
+      about the best soccer players"},{"role":"user","content":"\nCurrent Task: Top
+      10 best players in the world?\n\nProvide your complete response:"}],"model":"gpt-4.1-mini"}'
     headers:
+      User-Agent:
+      - X-USER-AGENT-XXX
       accept:
       - application/json
       accept-encoding:
-      - gzip, deflate, zstd
+      - ACCEPT-ENCODING-XXX
+      authorization:
+      - AUTHORIZATION-XXX
       connection:
       - keep-alive
       content-length:
-      - '694'
+      - '404'
       content-type:
       - application/json
       host:
       - api.openai.com
-      user-agent:
-      - OpenAI/Python 1.78.0
       x-stainless-arch:
-      - arm64
+      - X-STAINLESS-ARCH-XXX
       x-stainless-async:
       - 'false'
       x-stainless-lang:
       - python
       x-stainless-os:
-      - MacOS
+      - X-STAINLESS-OS-XXX
       x-stainless-package-version:
-      - 1.78.0
-      x-stainless-raw-response:
-      - 'true'
+      - 1.83.0
       x-stainless-read-timeout:
-      - '600.0'
+      - X-STAINLESS-READ-TIMEOUT-XXX
       x-stainless-retry-count:
       - '0'
       x-stainless-runtime:
       - CPython
       x-stainless-runtime-version:
-      - 3.12.9
+      - 3.13.5
     method: POST
     uri: https://api.openai.com/v1/chat/completions
   response:
     body:
-      string: "{\n  \"id\": \"chatcmpl-BgufUtDqGzvqPZx2NmkqqxdW4G8rQ\",\n  \"object\": \"chat.completion\",\n  \"created\": 1749567308,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n  \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\": \"assistant\",\n        \"content\": \"Thought: I now can give a great answer  \\nFinal Answer: The top 10 best soccer players in the world, as of October 2023, can be identified based on their recent performances, skills, impact on games, and overall contributions to their teams. Here is the structured list:\\n\\n1. **Lionel Messi (Inter Miami CF)**\\n   - Position: Forward\\n   - Key Attributes: Dribbling, vision, goal-scoring ability.\\n   - Achievements: Multiple Ballon d'Or winner, Copa America champion, World Cup champion (2022).\\n\\n2. **Kylian Mbappé (Paris Saint-Germain)**\\n   - Position: Forward\\n   - Key Attributes: Speed, technique, finishing.\\n   - Achievements: FIFA World Cup champion (2018), Ligue 1 titles, multiple\
-        \ domestic cups.\\n\\n3. **Erling Haaland (Manchester City)**\\n   - Position: Forward\\n   - Key Attributes: Power, speed, goal-scoring instinct.\\n   - Achievements: Bundesliga top scorer, UEFA Champions League winner (2023), Premier League titles.\\n\\n4. **Kevin De Bruyne (Manchester City)**\\n   - Position: Midfielder\\n   - Key Attributes: Passing, vision, creativity.\\n   - Achievements: Multiple Premier League titles, FA Cups, UEFA Champions League winner (2023).\\n\\n5. **Karim Benzema (Al-Ittihad)**\\n   - Position: Forward\\n   - Key Attributes: Goal-scoring, playmaking, tactical intelligence.\\n   - Achievements: 2022 Ballon d'Or winner, multiple Champions Leagues with Real Madrid.\\n\\n6. **Neymar Jr. (Al Hilal)**\\n   - Position: Forward\\n   - Key Attributes: Flair, dribbling, creativity.\\n   - Achievements: Multiple domestic league titles, Champions League runner-up.\\n\\n7. **Robert Lewandowski (FC Barcelona)**\\n   - Position: Forward\\n   - Key Attributes: Finishing,\
-        \ positioning, aerial ability.\\n   - Achievements: FIFA Best Men's Player, multiple Bundesliga titles, La Liga champion (2023).\\n\\n8. **Mohamed Salah (Liverpool)**\\n   - Position: Forward\\n   - Key Attributes: Speed, finishing, dribbling.\\n   - Achievements: Premier League champion, FA Cup, UEFA Champions League winner.\\n\\n9. **Vinícius Júnior (Real Madrid)**\\n   - Position: Forward\\n   - Key Attributes: Speed, dribbling, creativity.\\n   - Achievements: UEFA Champions League winner (2022), La Liga champion (2023).\\n\\n10. **Luka Modrić (Real Madrid)**\\n    - Position: Midfielder\\n    - Key Attributes: Passing, vision, tactical intelligence.\\n    - Achievements: Multiple Champions League titles, Ballon d'Or winner (2018).\\n\\nThis list is compiled based on their current form, past performances, and contributions to their respective teams in both domestic and international competitions. Player rankings can vary based on personal opinion and specific criteria used for\
-        \ evaluation, but these players have consistently been regarded as some of the best in the world as of October 2023.\",\n        \"refusal\": null,\n        \"annotations\": []\n      },\n      \"logprobs\": null,\n      \"finish_reason\": \"stop\"\n    }\n  ],\n  \"usage\": {\n    \"prompt_tokens\": 122,\n    \"completion_tokens\": 643,\n    \"total_tokens\": 765,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\": 0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\": {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\": 0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\": \"default\",\n  \"system_fingerprint\": \"fp_34a54ae93c\"\n}\n"
+      string: "{\n  \"id\": \"chatcmpl-D6L3hzoRVVEa07HZsM9wpi2RVRKQp\",\n  \"object\":
+        \"chat.completion\",\n  \"created\": 1770403289,\n  \"model\": \"gpt-4.1-mini-2025-04-14\",\n
+        \ \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\":
+        \"assistant\",\n        \"content\": \"Here is a structured list of the top
+        10 best soccer players in the world as of 2024, based on recent performances,
+        awards, and overall impact on the game:\\n\\n1. **Kylian Mbapp\xE9**  \\n
+        \  - Nationality: French  \\n   - Club: Paris Saint-Germain (PSG)  \\n   -
+        Position: Forward  \\n   - Key Highlights: Multiple Ligue 1 titles, World
+        Cup winner (2018), known for speed, dribbling, and scoring prowess.\\n\\n2.
+        **Erling Haaland**  \\n   - Nationality: Norwegian  \\n   - Club: Manchester
+        City  \\n   - Position: Striker  \\n   - Key Highlights: Premier League Golden
+        Boot winner, incredible goal-scoring record, physical presence, and finishing
+        skills.\\n\\n3. **Lionel Messi**  \\n   - Nationality: Argentine  \\n   -
+        Club: Inter Miami  \\n   - Position: Forward/Attacking Midfielder  \\n   -
+        Key Highlights: Seven Ballon d\u2019Or awards, World Cup winner (2022), exceptional
+        playmaking and dribbling ability.\\n\\n4. **Kevin De Bruyne**  \\n   - Nationality:
+        Belgian  \\n   - Club: Manchester City  \\n   - Position: Midfielder  \\n
+        \  - Key Highlights: One of the best playmakers globally, assists leader,
+        consistent high-level performance in the Premier League.\\n\\n5. **Robert
+        Lewandowski**  \\n   - Nationality: Polish  \\n   - Club: FC Barcelona  \\n
+        \  - Position: Striker  \\n   - Key Highlights: Exceptional goal-scoring record,
+        multiple Bundesliga top scorer awards, key figure in Bayern Munich\u2019s
+        dominance before transferring.\\n\\n6. **Karim Benzema**  \\n   - Nationality:
+        French  \\n   - Club: Al-Ittihad  \\n   - Position: Striker  \\n   - Key Highlights:
+        Ballon d\u2019Or winner (2022), excellent technical skills, leadership at
+        Real Madrid before recent transfer.\\n\\n7. **Mohamed Salah**  \\n   - Nationality:
+        Egyptian  \\n   - Club: Liverpool  \\n   - Position: Forward  \\n   - Key
+        Highlights: Premier League Golden Boot winner, known for speed, dribbling,
+        and goal-scoring consistency.\\n\\n8. **Vin\xEDcius J\xFAnior**  \\n   - Nationality:
+        Brazilian  \\n   - Club: Real Madrid  \\n   - Position: Winger  \\n   - Key
+        Highlights: Key player for Real Madrid, exceptional dribbling and pace, rising
+        star in world football.\\n\\n9. **Jude Bellingham**  \\n   - Nationality:
+        English  \\n   - Club: Real Madrid  \\n   - Position: Midfielder  \\n   -
+        Key Highlights: Young talent with maturity beyond years, influential midfielder
+        with great vision and work rate.\\n\\n10. **Thibaut Courtois**  \\n    - Nationality:
+        Belgian  \\n    - Club: Real Madrid  \\n    - Position: Goalkeeper  \\n    -
+        Key Highlights: One of the best goalkeepers globally, crucial performances
+        in La Liga and Champions League.\\n\\nThese rankings consider individual talent,
+        recent achievements, influence on matches, and overall contribution to club
+        and country.\",\n        \"refusal\": null,\n        \"annotations\": []\n
+        \     },\n      \"logprobs\": null,\n      \"finish_reason\": \"stop\"\n    }\n
+        \ ],\n  \"usage\": {\n    \"prompt_tokens\": 68,\n    \"completion_tokens\":
+        621,\n    \"total_tokens\": 689,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\":
+        0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\":
+        {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\":
+        0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\":
+        \"default\",\n  \"system_fingerprint\": \"fp_75546bd1a7\"\n}\n"
     headers:
       CF-RAY:
-      - 94d9b5400dcd624b-GRU
+      - CF-RAY-XXX
       Connection:
       - keep-alive
       Content-Type:
       - application/json
       Date:
-      - Tue, 10 Jun 2025 14:55:42 GMT
+      - Fri, 06 Feb 2026 18:41:40 GMT
       Server:
       - cloudflare
       Set-Cookie:
-      - __cf_bm=8Yv8F0ZCFAo2lf.qoqxao70yxyjVvIV90zQqVF6bVzQ-1749567342-1.0.1.1-fZgnv3RDfunvCO1koxwwFJrHnxSx_rwS_FHvQ6xxDPpKHwYr7dTqIQLZrNgSX5twGyK4F22rUmkuiS6KMVogcinChk8lmHtJBTUVTFjr2KU; path=/; expires=Tue, 10-Jun-25 15:25:42 GMT; domain=.api.openai.com; HttpOnly; Secure; SameSite=None
-      - _cfuvid=wzh8YnmXvLq1G0RcIVijtzboQtCZyIe2uZiochkBLqE-1749567342267-0.0.1.1-604800000; path=/; domain=.api.openai.com; HttpOnly; Secure; SameSite=None
+      - SET-COOKIE-XXX
+      Strict-Transport-Security:
+      - STS-XXX
       Transfer-Encoding:
       - chunked
       X-Content-Type-Options:
-      - nosniff
+      - X-CONTENT-TYPE-XXX
       access-control-expose-headers:
-      - X-Request-ID
+      - ACCESS-CONTROL-XXX
       alt-svc:
       - h3=":443"; ma=86400
       cf-cache-status:
       - DYNAMIC
       openai-organization:
-      - crewai-iuxna1
+      - OPENAI-ORG-XXX
       openai-processing-ms:
-      - '33288'
+      - '10634'
+      openai-project:
+      - OPENAI-PROJECT-XXX
       openai-version:
       - '2020-10-01'
-      strict-transport-security:
-      - max-age=31536000; includeSubDomains; preload
-      x-envoy-upstream-service-time:
-      - '33292'
+      x-openai-proxy-wasm:
+      - v0.1
       x-ratelimit-limit-requests:
-      - '30000'
+      - X-RATELIMIT-LIMIT-REQUESTS-XXX
       x-ratelimit-limit-tokens:
-      - '150000000'
+      - X-RATELIMIT-LIMIT-TOKENS-XXX
       x-ratelimit-remaining-requests:
-      - '29999'
+      - X-RATELIMIT-REMAINING-REQUESTS-XXX
       x-ratelimit-remaining-tokens:
-      - '149999859'
+      - X-RATELIMIT-REMAINING-TOKENS-XXX
       x-ratelimit-reset-requests:
-      - 2ms
+      - X-RATELIMIT-RESET-REQUESTS-XXX
       x-ratelimit-reset-tokens:
-      - 0s
+      - X-RATELIMIT-RESET-TOKENS-XXX
       x-request-id:
-      - req_6a587ea22edef774ecdada790a320cab
+      - X-REQUEST-ID-XXX
     status:
       code: 200
       message: OK
 - request:
-    body: '{"messages": [{"role": "system", "content": "You are Sports Analyst. You are an expert at gathering and organizing information. You carefully collect details and present them in a structured way.\nYour personal goal is: Gather information about the best soccer players\n\nTo give my best complete final answer to the task respond using the exact following format:\n\nThought: I now can give a great answer\nFinal Answer: Your final answer must be the great and the most complete as possible, it must be outcome described.\n\nI MUST use these formats, my job depends on it!"}, {"role": "user", "content": "Top 10 best players in the world?"}, {"role": "assistant", "content": "Thought: I now can give a great answer  \nFinal Answer: The top 10 best soccer players in the world, as of October 2023, can be identified based on their recent performances, skills, impact on games, and overall contributions to their teams. Here is the structured list:\n\n1. **Lionel Messi (Inter Miami CF)**\n   -
-      Position: Forward\n   - Key Attributes: Dribbling, vision, goal-scoring ability.\n   - Achievements: Multiple Ballon d''Or winner, Copa America champion, World Cup champion (2022).\n\n2. **Kylian Mbapp\u00e9 (Paris Saint-Germain)**\n   - Position: Forward\n   - Key Attributes: Speed, technique, finishing.\n   - Achievements: FIFA World Cup champion (2018), Ligue 1 titles, multiple domestic cups.\n\n3. **Erling Haaland (Manchester City)**\n   - Position: Forward\n   - Key Attributes: Power, speed, goal-scoring instinct.\n   - Achievements: Bundesliga top scorer, UEFA Champions League winner (2023), Premier League titles.\n\n4. **Kevin De Bruyne (Manchester City)**\n   - Position: Midfielder\n   - Key Attributes: Passing, vision, creativity.\n   - Achievements: Multiple Premier League titles, FA Cups, UEFA Champions League winner (2023).\n\n5. **Karim Benzema (Al-Ittihad)**\n   - Position: Forward\n   - Key Attributes: Goal-scoring, playmaking, tactical intelligence.\n   - Achievements:
-      2022 Ballon d''Or winner, multiple Champions Leagues with Real Madrid.\n\n6. **Neymar Jr. (Al Hilal)**\n   - Position: Forward\n   - Key Attributes: Flair, dribbling, creativity.\n   - Achievements: Multiple domestic league titles, Champions League runner-up.\n\n7. **Robert Lewandowski (FC Barcelona)**\n   - Position: Forward\n   - Key Attributes: Finishing, positioning, aerial ability.\n   - Achievements: FIFA Best Men''s Player, multiple Bundesliga titles, La Liga champion (2023).\n\n8. **Mohamed Salah (Liverpool)**\n   - Position: Forward\n   - Key Attributes: Speed, finishing, dribbling.\n   - Achievements: Premier League champion, FA Cup, UEFA Champions League winner.\n\n9. **Vin\u00edcius J\u00fanior (Real Madrid)**\n   - Position: Forward\n   - Key Attributes: Speed, dribbling, creativity.\n   - Achievements: UEFA Champions League winner (2022), La Liga champion (2023).\n\n10. **Luka Modri\u0107 (Real Madrid)**\n    - Position: Midfielder\n    - Key Attributes: Passing, vision,
-      tactical intelligence.\n    - Achievements: Multiple Champions League titles, Ballon d''Or winner (2018).\n\nThis list is compiled based on their current form, past performances, and contributions to their respective teams in both domestic and international competitions. Player rankings can vary based on personal opinion and specific criteria used for evaluation, but these players have consistently been regarded as some of the best in the world as of October 2023."}, {"role": "user", "content": "You are not allowed to include Brazilian players"}], "model": "gpt-4o-mini", "stop": ["\nObservation:"]}'
+    body: '{"messages":[{"role":"system","content":"You are Sports Analyst. You are
+      an expert at gathering and organizing information. You carefully collect details
+      and present them in a structured way.\nYour personal goal is: Gather information
+      about the best soccer players"},{"role":"user","content":"\nCurrent Task: Top
+      10 best players in the world?\n\nProvide your complete response:"}],"model":"gpt-4.1-mini"}'
     headers:
+      User-Agent:
+      - X-USER-AGENT-XXX
       accept:
       - application/json
       accept-encoding:
-      - gzip, deflate, zstd
+      - ACCEPT-ENCODING-XXX
+      authorization:
+      - AUTHORIZATION-XXX
       connection:
       - keep-alive
       content-length:
-      - '3594'
+      - '404'
       content-type:
       - application/json
       cookie:
-      - __cf_bm=8Yv8F0ZCFAo2lf.qoqxao70yxyjVvIV90zQqVF6bVzQ-1749567342-1.0.1.1-fZgnv3RDfunvCO1koxwwFJrHnxSx_rwS_FHvQ6xxDPpKHwYr7dTqIQLZrNgSX5twGyK4F22rUmkuiS6KMVogcinChk8lmHtJBTUVTFjr2KU; _cfuvid=wzh8YnmXvLq1G0RcIVijtzboQtCZyIe2uZiochkBLqE-1749567342267-0.0.1.1-604800000
+      - COOKIE-XXX
       host:
       - api.openai.com
-      user-agent:
-      - OpenAI/Python 1.78.0
       x-stainless-arch:
-      - arm64
+      - X-STAINLESS-ARCH-XXX
       x-stainless-async:
       - 'false'
       x-stainless-lang:
       - python
       x-stainless-os:
-      - MacOS
+      - X-STAINLESS-OS-XXX
       x-stainless-package-version:
-      - 1.78.0
-      x-stainless-raw-response:
-      - 'true'
+      - 1.83.0
       x-stainless-read-timeout:
-      - '600.0'
+      - X-STAINLESS-READ-TIMEOUT-XXX
       x-stainless-retry-count:
       - '0'
       x-stainless-runtime:
       - CPython
       x-stainless-runtime-version:
-      - 3.12.9
+      - 3.13.5
     method: POST
     uri: https://api.openai.com/v1/chat/completions
   response:
     body:
-      string: "{\n  \"id\": \"chatcmpl-BgugJkCDtB2EfvAMiIFK0reeLKFBl\",\n  \"object\": \"chat.completion\",\n  \"created\": 1749567359,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n  \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\": \"assistant\",\n        \"content\": \"Thought: I now can give a great answer  \\nFinal Answer: Here is an updated list of the top 10 best soccer players in the world as of October 2023, excluding Brazilian players:\\n\\n1. **Lionel Messi (Inter Miami CF)**\\n   - Position: Forward\\n   - Key Attributes: Dribbling, vision, goal-scoring ability.\\n   - Achievements: Multiple Ballon d'Or winner, Copa America champion, World Cup champion (2022).\\n\\n2. **Kylian Mbappé (Paris Saint-Germain)**\\n   - Position: Forward\\n   - Key Attributes: Speed, technique, finishing.\\n   - Achievements: FIFA World Cup champion (2018), Ligue 1 titles, multiple domestic cups.\\n\\n3. **Erling Haaland (Manchester City)**\\n   - Position: Forward\\\
-        n   - Key Attributes: Power, speed, goal-scoring instinct.\\n   - Achievements: Bundesliga top scorer, UEFA Champions League winner (2023), Premier League titles.\\n\\n4. **Kevin De Bruyne (Manchester City)**\\n   - Position: Midfielder\\n   - Key Attributes: Passing, vision, creativity.\\n   - Achievements: Multiple Premier League titles, FA Cups, UEFA Champions League winner (2023).\\n\\n5. **Karim Benzema (Al-Ittihad)**\\n   - Position: Forward\\n   - Key Attributes: Goal-scoring, playmaking, tactical intelligence.\\n   - Achievements: 2022 Ballon d'Or winner, multiple Champions Leagues with Real Madrid.\\n\\n6. **Robert Lewandowski (FC Barcelona)**\\n   - Position: Forward\\n   - Key Attributes: Finishing, positioning, aerial ability.\\n   - Achievements: FIFA Best Men's Player, multiple Bundesliga titles, La Liga champion (2023).\\n\\n7. **Mohamed Salah (Liverpool)**\\n   - Position: Forward\\n   - Key Attributes: Speed, finishing, dribbling.\\n   - Achievements: Premier League\
-        \ champion, FA Cup, UEFA Champions League winner.\\n\\n8. **Vinícius Júnior (Real Madrid)**\\n   - Position: Forward\\n   - Key Attributes: Speed, dribbling, creativity.\\n   - Achievements: UEFA Champions League winner (2022), La Liga champion (2023).\\n\\n9. **Luka Modrić (Real Madrid)**\\n   - Position: Midfielder\\n   - Key Attributes: Passing, vision, tactical intelligence.\\n   - Achievements: Multiple Champions League titles, Ballon d'Or winner (2018).\\n\\n10. **Harry Kane (Bayern Munich)**\\n    - Position: Forward\\n    - Key Attributes: Goal-scoring, technique, playmaking.\\n    - Achievements: Golden Boot winner, Premier League titles, UEFA European Championship runner-up.\\n\\nThis list has been adjusted to exclude Brazilian players and focuses on those who have made significant impacts in their clubs and on the international stage as of October 2023. Each player is recognized for their exceptional skills, performances, and achievements.\",\n        \"refusal\": null,\n\
-        \        \"annotations\": []\n      },\n      \"logprobs\": null,\n      \"finish_reason\": \"stop\"\n    }\n  ],\n  \"usage\": {\n    \"prompt_tokens\": 781,\n    \"completion_tokens\": 610,\n    \"total_tokens\": 1391,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\": 0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\": {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\": 0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\": \"default\",\n  \"system_fingerprint\": \"fp_34a54ae93c\"\n}\n"
+      string: "{\n  \"id\": \"chatcmpl-D6L3sn9nSnGGOMKrS88avliVF7XTv\",\n  \"object\":
+        \"chat.completion\",\n  \"created\": 1770403300,\n  \"model\": \"gpt-4.1-mini-2025-04-14\",\n
+        \ \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\":
+        \"assistant\",\n        \"content\": \"Certainly! Here's a structured list
+        of the top 10 best soccer players in the world as of 2024, considering their
+        performance, skills, achievements, and impact in recent seasons:\\n\\n###
+        Top 10 Best Soccer Players in the World (2024)\\n\\n| Rank | Player Name         |
+        Nationality | Club (2023/24 Season)      | Position         | Key Attributes
+        \                  | Recent Achievements                  |\\n|-------|---------------------|-------------|----------------------------|------------------|---------------------------------|------------------------------------|\\n|
+        1     | Lionel Messi        | Argentina   | Paris Saint-Germain (PSG)   |
+        Forward/Playmaker| Dribbling, Vision, Free kicks   | 2023 World Cup Golden
+        Ball, Club Successes |\\n| 2     | Kylian Mbapp\xE9       | France      |
+        Paris Saint-Germain (PSG)   | Forward          | Speed, Finishing, Dribbling
+        \    | Ligue 1 Top Scorer, World Cup Winner 2018|\\n| 3     | Erling Haaland
+        \     | Norway      | Manchester City             | Striker          | Strength,
+        Finishing, Positioning| Premier League Golden Boot, Champions League Impact|\\n|
+        4     | Kevin De Bruyne     | Belgium     | Manchester City             |
+        Midfielder       | Passing, Vision, Creativity     | Premier League Titles,
+        Key Playmaker|\\n| 5     | Robert Lewandowski  | Poland      | FC Barcelona
+        \              | Striker          | Finishing, Positioning, Composure| La
+        Liga Top Scorer, Consistent Scorer|\\n| 6     | Neymar Jr.          | Brazil
+        \     | Al-Hilal                   | Forward/Winger   | Dribbling, Creativity,
+        Flair    | Copa America Titles, Club Success |\\n| 7     | Mohamed Salah       |
+        Egypt       | Liverpool                  | Forward/Winger   | Pace, Finishing,
+        Work Rate      | Premier League Golden Boot, Champions League Winner|\\n|
+        8     | Vin\xEDcius Jr.        | Brazil      | Real Madrid                |
+        Winger           | Speed, Dribbling, Crossing     | La Liga Titles, UEFA Champions
+        League Winner|\\n| 9     | Luka Modri\u0107         | Croatia     | Real Madrid
+        \               | Midfielder       | Passing, Control, Experience    | Ballon
+        d\u2019Or 2018, Multiple Champions League Titles|\\n| 10    | Karim Benzema
+        \      | France      | Al-Ittihad                | Striker          | Finishing,
+        Link-up Play, Movements| Ballon d\u2019Or 2022, UEFA Champions League Top
+        Scorer |\\n\\n### Notes:\\n- The rankings reflect a combination of individual
+        skill, recent performance, consistency, and influence on the game.\\n- Players\u2019
+        clubs are based on the 2023/24 season affiliations.\\n- Achievements highlight
+        recent titles, awards, or standout contributions.\\n\\nIf you would like me
+        to focus on specific leagues, historical players, or emerging talents, just
+        let me know!\",\n        \"refusal\": null,\n        \"annotations\": []\n
+        \     },\n      \"logprobs\": null,\n      \"finish_reason\": \"stop\"\n    }\n
+        \ ],\n  \"usage\": {\n    \"prompt_tokens\": 68,\n    \"completion_tokens\":
+        605,\n    \"total_tokens\": 673,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\":
+        0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\":
+        {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\":
+        0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\":
+        \"default\",\n  \"system_fingerprint\": \"fp_75546bd1a7\"\n}\n"
     headers:
       CF-RAY:
-      - 94d9b6782db84d3b-GRU
+      - CF-RAY-XXX
       Connection:
       - keep-alive
       Content-Type:
       - application/json
       Date:
-      - Tue, 10 Jun 2025 14:56:30 GMT
+      - Fri, 06 Feb 2026 18:41:49 GMT
       Server:
       - cloudflare
+      Strict-Transport-Security:
+      - STS-XXX
       Transfer-Encoding:
       - chunked
       X-Content-Type-Options:
-      - nosniff
+      - X-CONTENT-TYPE-XXX
       access-control-expose-headers:
-      - X-Request-ID
+      - ACCESS-CONTROL-XXX
       alt-svc:
       - h3=":443"; ma=86400
       cf-cache-status:
       - DYNAMIC
       openai-organization:
-      - crewai-iuxna1
+      - OPENAI-ORG-XXX
       openai-processing-ms:
-      - '31484'
+      - '9044'
+      openai-project:
+      - OPENAI-PROJECT-XXX
       openai-version:
       - '2020-10-01'
-      strict-transport-security:
-      - max-age=31536000; includeSubDomains; preload
-      x-envoy-upstream-service-time:
-      - '31490'
+      x-openai-proxy-wasm:
+      - v0.1
       x-ratelimit-limit-requests:
-      - '30000'
+      - X-RATELIMIT-LIMIT-REQUESTS-XXX
       x-ratelimit-limit-tokens:
-      - '150000000'
+      - X-RATELIMIT-LIMIT-TOKENS-XXX
       x-ratelimit-remaining-requests:
-      - '29999'
+      - X-RATELIMIT-REMAINING-REQUESTS-XXX
       x-ratelimit-remaining-tokens:
-      - '149999166'
+      - X-RATELIMIT-REMAINING-TOKENS-XXX
       x-ratelimit-reset-requests:
-      - 2ms
+      - X-RATELIMIT-RESET-REQUESTS-XXX
       x-ratelimit-reset-tokens:
-      - 0s
+      - X-RATELIMIT-RESET-TOKENS-XXX
       x-request-id:
-      - req_aa737cf40bb76af9f458bfd35f7a77a1
+      - X-REQUEST-ID-XXX
     status:
       code: 200
       message: OK
 - request:
-    body: '{"messages": [{"role": "system", "content": "You are Sports Analyst. You are an expert at gathering and organizing information. You carefully collect details and present them in a structured way.\nYour personal goal is: Gather information about the best soccer players\n\nTo give my best complete final answer to the task respond using the exact following format:\n\nThought: I now can give a great answer\nFinal Answer: Your final answer must be the great and the most complete as possible, it must be outcome described.\n\nI MUST use these formats, my job depends on it!"}, {"role": "user", "content": "Top 10 best players in the world?"}, {"role": "assistant", "content": "Thought: I now can give a great answer  \nFinal Answer: The top 10 best soccer players in the world, as of October 2023, can be identified based on their recent performances, skills, impact on games, and overall contributions to their teams. Here is the structured list:\n\n1. **Lionel Messi (Inter Miami CF)**\n   -
-      Position: Forward\n   - Key Attributes: Dribbling, vision, goal-scoring ability.\n   - Achievements: Multiple Ballon d''Or winner, Copa America champion, World Cup champion (2022).\n\n2. **Kylian Mbapp\u00e9 (Paris Saint-Germain)**\n   - Position: Forward\n   - Key Attributes: Speed, technique, finishing.\n   - Achievements: FIFA World Cup champion (2018), Ligue 1 titles, multiple domestic cups.\n\n3. **Erling Haaland (Manchester City)**\n   - Position: Forward\n   - Key Attributes: Power, speed, goal-scoring instinct.\n   - Achievements: Bundesliga top scorer, UEFA Champions League winner (2023), Premier League titles.\n\n4. **Kevin De Bruyne (Manchester City)**\n   - Position: Midfielder\n   - Key Attributes: Passing, vision, creativity.\n   - Achievements: Multiple Premier League titles, FA Cups, UEFA Champions League winner (2023).\n\n5. **Karim Benzema (Al-Ittihad)**\n   - Position: Forward\n   - Key Attributes: Goal-scoring, playmaking, tactical intelligence.\n   - Achievements:
-      2022 Ballon d''Or winner, multiple Champions Leagues with Real Madrid.\n\n6. **Neymar Jr. (Al Hilal)**\n   - Position: Forward\n   - Key Attributes: Flair, dribbling, creativity.\n   - Achievements: Multiple domestic league titles, Champions League runner-up.\n\n7. **Robert Lewandowski (FC Barcelona)**\n   - Position: Forward\n   - Key Attributes: Finishing, positioning, aerial ability.\n   - Achievements: FIFA Best Men''s Player, multiple Bundesliga titles, La Liga champion (2023).\n\n8. **Mohamed Salah (Liverpool)**\n   - Position: Forward\n   - Key Attributes: Speed, finishing, dribbling.\n   - Achievements: Premier League champion, FA Cup, UEFA Champions League winner.\n\n9. **Vin\u00edcius J\u00fanior (Real Madrid)**\n   - Position: Forward\n   - Key Attributes: Speed, dribbling, creativity.\n   - Achievements: UEFA Champions League winner (2022), La Liga champion (2023).\n\n10. **Luka Modri\u0107 (Real Madrid)**\n    - Position: Midfielder\n    - Key Attributes: Passing, vision,
-      tactical intelligence.\n    - Achievements: Multiple Champions League titles, Ballon d''Or winner (2018).\n\nThis list is compiled based on their current form, past performances, and contributions to their respective teams in both domestic and international competitions. Player rankings can vary based on personal opinion and specific criteria used for evaluation, but these players have consistently been regarded as some of the best in the world as of October 2023."}, {"role": "user", "content": "You are not allowed to include Brazilian players"}, {"role": "assistant", "content": "Thought: I now can give a great answer  \nFinal Answer: Here is an updated list of the top 10 best soccer players in the world as of October 2023, excluding Brazilian players:\n\n1. **Lionel Messi (Inter Miami CF)**\n   - Position: Forward\n   - Key Attributes: Dribbling, vision, goal-scoring ability.\n   - Achievements: Multiple Ballon d''Or winner, Copa America champion, World Cup champion (2022).\n\n2.
-      **Kylian Mbapp\u00e9 (Paris Saint-Germain)**\n   - Position: Forward\n   - Key Attributes: Speed, technique, finishing.\n   - Achievements: FIFA World Cup champion (2018), Ligue 1 titles, multiple domestic cups.\n\n3. **Erling Haaland (Manchester City)**\n   - Position: Forward\n   - Key Attributes: Power, speed, goal-scoring instinct.\n   - Achievements: Bundesliga top scorer, UEFA Champions League winner (2023), Premier League titles.\n\n4. **Kevin De Bruyne (Manchester City)**\n   - Position: Midfielder\n   - Key Attributes: Passing, vision, creativity.\n   - Achievements: Multiple Premier League titles, FA Cups, UEFA Champions League winner (2023).\n\n5. **Karim Benzema (Al-Ittihad)**\n   - Position: Forward\n   - Key Attributes: Goal-scoring, playmaking, tactical intelligence.\n   - Achievements: 2022 Ballon d''Or winner, multiple Champions Leagues with Real Madrid.\n\n6. **Robert Lewandowski (FC Barcelona)**\n   - Position: Forward\n   - Key Attributes: Finishing, positioning,
-      aerial ability.\n   - Achievements: FIFA Best Men''s Player, multiple Bundesliga titles, La Liga champion (2023).\n\n7. **Mohamed Salah (Liverpool)**\n   - Position: Forward\n   - Key Attributes: Speed, finishing, dribbling.\n   - Achievements: Premier League champion, FA Cup, UEFA Champions League winner.\n\n8. **Vin\u00edcius J\u00fanior (Real Madrid)**\n   - Position: Forward\n   - Key Attributes: Speed, dribbling, creativity.\n   - Achievements: UEFA Champions League winner (2022), La Liga champion (2023).\n\n9. **Luka Modri\u0107 (Real Madrid)**\n   - Position: Midfielder\n   - Key Attributes: Passing, vision, tactical intelligence.\n   - Achievements: Multiple Champions League titles, Ballon d''Or winner (2018).\n\n10. **Harry Kane (Bayern Munich)**\n    - Position: Forward\n    - Key Attributes: Goal-scoring, technique, playmaking.\n    - Achievements: Golden Boot winner, Premier League titles, UEFA European Championship runner-up.\n\nThis list has been adjusted to exclude Brazilian
-      players and focuses on those who have made significant impacts in their clubs and on the international stage as of October 2023. Each player is recognized for their exceptional skills, performances, and achievements."}, {"role": "user", "content": "You are not allowed to include Brazilian players"}], "model": "gpt-4o-mini", "stop": ["\nObservation:"]}'
+    body: '{"messages":[{"role":"system","content":"You are Sports Analyst. You are
+      an expert at gathering and organizing information. You carefully collect details
+      and present them in a structured way.\nYour personal goal is: Gather information
+      about the best soccer players"},{"role":"user","content":"\nCurrent Task: Top
+      10 best players in the world?\n\nProvide your complete response:"}],"model":"gpt-4.1-mini"}'
     headers:
+      User-Agent:
+      - X-USER-AGENT-XXX
       accept:
       - application/json
       accept-encoding:
-      - gzip, deflate, zstd
+      - ACCEPT-ENCODING-XXX
+      authorization:
+      - AUTHORIZATION-XXX
       connection:
       - keep-alive
       content-length:
-      - '6337'
+      - '404'
       content-type:
       - application/json
       cookie:
-      - __cf_bm=8Yv8F0ZCFAo2lf.qoqxao70yxyjVvIV90zQqVF6bVzQ-1749567342-1.0.1.1-fZgnv3RDfunvCO1koxwwFJrHnxSx_rwS_FHvQ6xxDPpKHwYr7dTqIQLZrNgSX5twGyK4F22rUmkuiS6KMVogcinChk8lmHtJBTUVTFjr2KU; _cfuvid=wzh8YnmXvLq1G0RcIVijtzboQtCZyIe2uZiochkBLqE-1749567342267-0.0.1.1-604800000
+      - COOKIE-XXX
       host:
       - api.openai.com
-      user-agent:
-      - OpenAI/Python 1.78.0
       x-stainless-arch:
-      - arm64
+      - X-STAINLESS-ARCH-XXX
       x-stainless-async:
       - 'false'
       x-stainless-lang:
       - python
       x-stainless-os:
-      - MacOS
+      - X-STAINLESS-OS-XXX
       x-stainless-package-version:
-      - 1.78.0
-      x-stainless-raw-response:
-      - 'true'
+      - 1.83.0
       x-stainless-read-timeout:
-      - '600.0'
+      - X-STAINLESS-READ-TIMEOUT-XXX
       x-stainless-retry-count:
       - '0'
       x-stainless-runtime:
       - CPython
       x-stainless-runtime-version:
-      - 3.12.9
+      - 3.13.5
     method: POST
     uri: https://api.openai.com/v1/chat/completions
   response:
     body:
-      string: "{\n  \"id\": \"chatcmpl-BgugsAmyI50uQ6SpCp89ZZY4eD1Pz\",\n  \"object\": \"chat.completion\",\n  \"created\": 1749567394,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n  \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\": \"assistant\",\n        \"content\": \"Thought: I now can give a great answer  \\nFinal Answer: Here is the revised list of the top 10 best soccer players in the world as of October 2023, explicitly excluding Brazilian players:\\n\\n1. **Lionel Messi (Inter Miami CF)**\\n   - Position: Forward\\n   - Key Attributes: Dribbling, vision, goal-scoring ability.\\n   - Achievements: Multiple Ballon d'Or winner, Copa America champion, World Cup champion (2022).\\n\\n2. **Kylian Mbappé (Paris Saint-Germain)**\\n   - Position: Forward\\n   - Key Attributes: Speed, technique, finishing.\\n   - Achievements: FIFA World Cup champion (2018), multiple Ligue 1 titles, and various domestic cups.\\n\\n3. **Erling Haaland (Manchester City)**\\n\
-        \   - Position: Forward\\n   - Key Attributes: Power, speed, goal-scoring instinct.\\n   - Achievements: Bundesliga top scorer, UEFA Champions League winner (2023), Premier League titles.\\n\\n4. **Kevin De Bruyne (Manchester City)**\\n   - Position: Midfielder\\n   - Key Attributes: Passing, vision, creativity.\\n   - Achievements: Multiple Premier League titles, FA Cups, UEFA Champions League winner (2023).\\n\\n5. **Karim Benzema (Al-Ittihad)**\\n   - Position: Forward\\n   - Key Attributes: Goal-scoring, playmaking, tactical intelligence.\\n   - Achievements: 2022 Ballon d'Or winner, multiple Champions Leagues with Real Madrid.\\n\\n6. **Robert Lewandowski (FC Barcelona)**\\n   - Position: Forward\\n   - Key Attributes: Finishing, positioning, aerial ability.\\n   - Achievements: FIFA Best Men's Player, multiple Bundesliga titles, La Liga champion (2023).\\n\\n7. **Mohamed Salah (Liverpool)**\\n   - Position: Forward\\n   - Key Attributes: Speed, finishing, dribbling.\\n   -\
-        \ Achievements: Premier League champion, FA Cup, UEFA Champions League winner.\\n\\n8. **Luka Modrić (Real Madrid)**\\n   - Position: Midfielder\\n   - Key Attributes: Passing, vision, tactical intelligence.\\n   - Achievements: Multiple Champions League titles, Ballon d'Or winner (2018).\\n\\n9. **Harry Kane (Bayern Munich)**\\n   - Position: Forward\\n   - Key Attributes: Goal-scoring, technique, playmaking.\\n   - Achievements: Golden Boot winner, Premier League titles, UEFA European Championship runner-up.\\n\\n10. **Rodri (Manchester City)**\\n    - Position: Midfielder\\n    - Key Attributes: Defensive skills, passing, positional awareness.\\n    - Achievements: Premier League titles, UEFA Champions League winner (2023).\\n\\nThis list is curated while adhering to the restriction of excluding Brazilian players. Each player included has demonstrated exceptional skills and remarkable performances, solidifying their status as some of the best in the world as of October 2023.\"\
-        ,\n        \"refusal\": null,\n        \"annotations\": []\n      },\n      \"logprobs\": null,\n      \"finish_reason\": \"stop\"\n    }\n  ],\n  \"usage\": {\n    \"prompt_tokens\": 1407,\n    \"completion_tokens\": 605,\n    \"total_tokens\": 2012,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\": 0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\": {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\": 0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\": \"default\",\n  \"system_fingerprint\": \"fp_62a23a81ef\"\n}\n"
+      string: "{\n  \"id\": \"chatcmpl-D6L4102eMwTEPeHxfyN9Kh7rjBoX6\",\n  \"object\":
+        \"chat.completion\",\n  \"created\": 1770403309,\n  \"model\": \"gpt-4.1-mini-2025-04-14\",\n
+        \ \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\":
+        \"assistant\",\n        \"content\": \"Certainly! Here is a list of the top
+        10 best soccer players in the world as of 2024, considering their recent performances,
+        skills, impact, and accolades:\\n\\n1. **Lionel Messi**  \\n   - Nationality:
+        Argentine  \\n   - Position: Forward  \\n   - Key Achievements: 7 Ballon d'Or
+        awards, led Argentina to 2021 Copa Am\xE9rica victory and 2022 FIFA World
+        Cup triumph, exceptional dribbling and playmaking skills.\\n\\n2. **Kylian
+        Mbapp\xE9**  \\n   - Nationality: French  \\n   - Position: Forward  \\n   -
+        Key Achievements: FIFA World Cup winner (2018), multiple Ligue 1 titles, known
+        for incredible speed, finishing, and consistency.\\n\\n3. **Erling Haaland**
+        \ \\n   - Nationality: Norwegian  \\n   - Position: Striker  \\n   - Key Achievements:
+        Premier League Golden Boot winner (2022-23), prolific goal scorer, physical
+        presence, and finishing ability.\\n\\n4. **Karim Benzema**  \\n   - Nationality:
+        French  \\n   - Position: Forward  \\n   - Key Achievements: 2022 Ballon d'Or
+        winner, key player for Real Madrid\u2019s recent Champions League victories,
+        excellent technical skills and leadership.\\n\\n5. **Kevin De Bruyne**  \\n
+        \  - Nationality: Belgian  \\n   - Position: Midfielder  \\n   - Key Achievements:
+        Premier League playmaker, known for vision, passing accuracy, and creativity.\\n\\n6.
+        **Robert Lewandowski**  \\n   - Nationality: Polish  \\n   - Position: Striker
+        \ \\n   - Key Achievements: Multiple Bundesliga top scorer titles, consistent
+        goal scorer, known for positioning and finishing.\\n\\n7. **Neymar Jr.**  \\n
+        \  - Nationality: Brazilian  \\n   - Position: Forward  \\n   - Key Achievements:
+        Exceptional dribbling, creativity, and flair; multiple domestic titles and
+        Copa Libertadores winner.\\n\\n8. **Mohamed Salah**  \\n   - Nationality:
+        Egyptian  \\n   - Position: Forward  \\n   - Key Achievements: Premier League
+        Golden Boot, consistent goal scoring with Liverpool, known for speed and finishing.\\n\\n9.
+        **Luka Modri\u0107**  \\n   - Nationality: Croatian  \\n   - Position: Midfielder
+        \ \\n   - Key Achievements: 2018 Ballon d\u2019Or winner, pivotal midfield
+        maestro, excellent passing and control.\\n\\n10. **Thibaut Courtois**  \\n
+        \   - Nationality: Belgian  \\n    - Position: Goalkeeper  \\n    - Key Achievements:
+        Exceptional shot-stopper, key player in Real Madrid's recent successes.\\n\\nThis
+        list includes a blend of forwards, midfielders, and a goalkeeper, showcasing
+        the best talents in various positions worldwide. The rankings may vary slightly
+        depending on current form and opinions, but these players consistently rank
+        among the best globally.\",\n        \"refusal\": null,\n        \"annotations\":
+        []\n      },\n      \"logprobs\": null,\n      \"finish_reason\": \"stop\"\n
+        \   }\n  ],\n  \"usage\": {\n    \"prompt_tokens\": 68,\n    \"completion_tokens\":
+        575,\n    \"total_tokens\": 643,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\":
+        0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\":
+        {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\":
+        0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\":
+        \"default\",\n  \"system_fingerprint\": \"fp_75546bd1a7\"\n}\n"
     headers:
       CF-RAY:
-      - 94d9b7561f204d3b-GRU
+      - CF-RAY-XXX
       Connection:
       - keep-alive
       Content-Type:
       - application/json
       Date:
-      - Tue, 10 Jun 2025 14:56:46 GMT
+      - Fri, 06 Feb 2026 18:41:57 GMT
       Server:
       - cloudflare
+      Strict-Transport-Security:
+      - STS-XXX
       Transfer-Encoding:
       - chunked
       X-Content-Type-Options:
-      - nosniff
+      - X-CONTENT-TYPE-XXX
       access-control-expose-headers:
-      - X-Request-ID
+      - ACCESS-CONTROL-XXX
       alt-svc:
       - h3=":443"; ma=86400
       cf-cache-status:
       - DYNAMIC
       openai-organization:
-      - crewai-iuxna1
+      - OPENAI-ORG-XXX
       openai-processing-ms:
-      - '12189'
+      - '7948'
+      openai-project:
+      - OPENAI-PROJECT-XXX
       openai-version:
       - '2020-10-01'
-      strict-transport-security:
-      - max-age=31536000; includeSubDomains; preload
-      x-envoy-upstream-service-time:
-      - '12193'
+      x-openai-proxy-wasm:
+      - v0.1
       x-ratelimit-limit-requests:
-      - '30000'
+      - X-RATELIMIT-LIMIT-REQUESTS-XXX
       x-ratelimit-limit-tokens:
-      - '150000000'
+      - X-RATELIMIT-LIMIT-TOKENS-XXX
       x-ratelimit-remaining-requests:
-      - '29999'
+      - X-RATELIMIT-REMAINING-REQUESTS-XXX
       x-ratelimit-remaining-tokens:
-      - '149998513'
+      - X-RATELIMIT-REMAINING-TOKENS-XXX
       x-ratelimit-reset-requests:
-      - 2ms
+      - X-RATELIMIT-RESET-REQUESTS-XXX
       x-ratelimit-reset-tokens:
-      - 0s
+      - X-RATELIMIT-RESET-TOKENS-XXX
       x-request-id:
-      - req_1098f5a5384f4a26aecf0c9e4e4d1fc0
-    status:
-      code: 200
-      message: OK
-- request:
-    body: '{"messages": [{"role": "system", "content": "You are Sports Analyst. You are an expert at gathering and organizing information. You carefully collect details and present them in a structured way.\nYour personal goal is: Gather information about the best soccer players\n\nTo give my best complete final answer to the task respond using the exact following format:\n\nThought: I now can give a great answer\nFinal Answer: Your final answer must be the great and the most complete as possible, it must be outcome described.\n\nI MUST use these formats, my job depends on it!"}, {"role": "user", "content": "Top 10 best players in the world?"}, {"role": "assistant", "content": "Thought: I now can give a great answer  \nFinal Answer: The top 10 best soccer players in the world, as of October 2023, can be identified based on their recent performances, skills, impact on games, and overall contributions to their teams. Here is the structured list:\n\n1. **Lionel Messi (Inter Miami CF)**\n   -
-      Position: Forward\n   - Key Attributes: Dribbling, vision, goal-scoring ability.\n   - Achievements: Multiple Ballon d''Or winner, Copa America champion, World Cup champion (2022).\n\n2. **Kylian Mbapp\u00e9 (Paris Saint-Germain)**\n   - Position: Forward\n   - Key Attributes: Speed, technique, finishing.\n   - Achievements: FIFA World Cup champion (2018), Ligue 1 titles, multiple domestic cups.\n\n3. **Erling Haaland (Manchester City)**\n   - Position: Forward\n   - Key Attributes: Power, speed, goal-scoring instinct.\n   - Achievements: Bundesliga top scorer, UEFA Champions League winner (2023), Premier League titles.\n\n4. **Kevin De Bruyne (Manchester City)**\n   - Position: Midfielder\n   - Key Attributes: Passing, vision, creativity.\n   - Achievements: Multiple Premier League titles, FA Cups, UEFA Champions League winner (2023).\n\n5. **Karim Benzema (Al-Ittihad)**\n   - Position: Forward\n   - Key Attributes: Goal-scoring, playmaking, tactical intelligence.\n   - Achievements:
-      2022 Ballon d''Or winner, multiple Champions Leagues with Real Madrid.\n\n6. **Neymar Jr. (Al Hilal)**\n   - Position: Forward\n   - Key Attributes: Flair, dribbling, creativity.\n   - Achievements: Multiple domestic league titles, Champions League runner-up.\n\n7. **Robert Lewandowski (FC Barcelona)**\n   - Position: Forward\n   - Key Attributes: Finishing, positioning, aerial ability.\n   - Achievements: FIFA Best Men''s Player, multiple Bundesliga titles, La Liga champion (2023).\n\n8. **Mohamed Salah (Liverpool)**\n   - Position: Forward\n   - Key Attributes: Speed, finishing, dribbling.\n   - Achievements: Premier League champion, FA Cup, UEFA Champions League winner.\n\n9. **Vin\u00edcius J\u00fanior (Real Madrid)**\n   - Position: Forward\n   - Key Attributes: Speed, dribbling, creativity.\n   - Achievements: UEFA Champions League winner (2022), La Liga champion (2023).\n\n10. **Luka Modri\u0107 (Real Madrid)**\n    - Position: Midfielder\n    - Key Attributes: Passing, vision,
-      tactical intelligence.\n    - Achievements: Multiple Champions League titles, Ballon d''Or winner (2018).\n\nThis list is compiled based on their current form, past performances, and contributions to their respective teams in both domestic and international competitions. Player rankings can vary based on personal opinion and specific criteria used for evaluation, but these players have consistently been regarded as some of the best in the world as of October 2023."}, {"role": "user", "content": "You are not allowed to include Brazilian players"}, {"role": "assistant", "content": "Thought: I now can give a great answer  \nFinal Answer: Here is an updated list of the top 10 best soccer players in the world as of October 2023, excluding Brazilian players:\n\n1. **Lionel Messi (Inter Miami CF)**\n   - Position: Forward\n   - Key Attributes: Dribbling, vision, goal-scoring ability.\n   - Achievements: Multiple Ballon d''Or winner, Copa America champion, World Cup champion (2022).\n\n2.
-      **Kylian Mbapp\u00e9 (Paris Saint-Germain)**\n   - Position: Forward\n   - Key Attributes: Speed, technique, finishing.\n   - Achievements: FIFA World Cup champion (2018), Ligue 1 titles, multiple domestic cups.\n\n3. **Erling Haaland (Manchester City)**\n   - Position: Forward\n   - Key Attributes: Power, speed, goal-scoring instinct.\n   - Achievements: Bundesliga top scorer, UEFA Champions League winner (2023), Premier League titles.\n\n4. **Kevin De Bruyne (Manchester City)**\n   - Position: Midfielder\n   - Key Attributes: Passing, vision, creativity.\n   - Achievements: Multiple Premier League titles, FA Cups, UEFA Champions League winner (2023).\n\n5. **Karim Benzema (Al-Ittihad)**\n   - Position: Forward\n   - Key Attributes: Goal-scoring, playmaking, tactical intelligence.\n   - Achievements: 2022 Ballon d''Or winner, multiple Champions Leagues with Real Madrid.\n\n6. **Robert Lewandowski (FC Barcelona)**\n   - Position: Forward\n   - Key Attributes: Finishing, positioning,
-      aerial ability.\n   - Achievements: FIFA Best Men''s Player, multiple Bundesliga titles, La Liga champion (2023).\n\n7. **Mohamed Salah (Liverpool)**\n   - Position: Forward\n   - Key Attributes: Speed, finishing, dribbling.\n   - Achievements: Premier League champion, FA Cup, UEFA Champions League winner.\n\n8. **Vin\u00edcius J\u00fanior (Real Madrid)**\n   - Position: Forward\n   - Key Attributes: Speed, dribbling, creativity.\n   - Achievements: UEFA Champions League winner (2022), La Liga champion (2023).\n\n9. **Luka Modri\u0107 (Real Madrid)**\n   - Position: Midfielder\n   - Key Attributes: Passing, vision, tactical intelligence.\n   - Achievements: Multiple Champions League titles, Ballon d''Or winner (2018).\n\n10. **Harry Kane (Bayern Munich)**\n    - Position: Forward\n    - Key Attributes: Goal-scoring, technique, playmaking.\n    - Achievements: Golden Boot winner, Premier League titles, UEFA European Championship runner-up.\n\nThis list has been adjusted to exclude Brazilian
-      players and focuses on those who have made significant impacts in their clubs and on the international stage as of October 2023. Each player is recognized for their exceptional skills, performances, and achievements."}, {"role": "user", "content": "You are not allowed to include Brazilian players"}, {"role": "assistant", "content": "Thought: I now can give a great answer  \nFinal Answer: Here is the revised list of the top 10 best soccer players in the world as of October 2023, explicitly excluding Brazilian players:\n\n1. **Lionel Messi (Inter Miami CF)**\n   - Position: Forward\n   - Key Attributes: Dribbling, vision, goal-scoring ability.\n   - Achievements: Multiple Ballon d''Or winner, Copa America champion, World Cup champion (2022).\n\n2. **Kylian Mbapp\u00e9 (Paris Saint-Germain)**\n   - Position: Forward\n   - Key Attributes: Speed, technique, finishing.\n   - Achievements: FIFA World Cup champion (2018), multiple Ligue 1 titles, and various domestic cups.\n\n3. **Erling Haaland
-      (Manchester City)**\n   - Position: Forward\n   - Key Attributes: Power, speed, goal-scoring instinct.\n   - Achievements: Bundesliga top scorer, UEFA Champions League winner (2023), Premier League titles.\n\n4. **Kevin De Bruyne (Manchester City)**\n   - Position: Midfielder\n   - Key Attributes: Passing, vision, creativity.\n   - Achievements: Multiple Premier League titles, FA Cups, UEFA Champions League winner (2023).\n\n5. **Karim Benzema (Al-Ittihad)**\n   - Position: Forward\n   - Key Attributes: Goal-scoring, playmaking, tactical intelligence.\n   - Achievements: 2022 Ballon d''Or winner, multiple Champions Leagues with Real Madrid.\n\n6. **Robert Lewandowski (FC Barcelona)**\n   - Position: Forward\n   - Key Attributes: Finishing, positioning, aerial ability.\n   - Achievements: FIFA Best Men''s Player, multiple Bundesliga titles, La Liga champion (2023).\n\n7. **Mohamed Salah (Liverpool)**\n   - Position: Forward\n   - Key Attributes: Speed, finishing, dribbling.\n   - Achievements:
-      Premier League champion, FA Cup, UEFA Champions League winner.\n\n8. **Luka Modri\u0107 (Real Madrid)**\n   - Position: Midfielder\n   - Key Attributes: Passing, vision, tactical intelligence.\n   - Achievements: Multiple Champions League titles, Ballon d''Or winner (2018).\n\n9. **Harry Kane (Bayern Munich)**\n   - Position: Forward\n   - Key Attributes: Goal-scoring, technique, playmaking.\n   - Achievements: Golden Boot winner, Premier League titles, UEFA European Championship runner-up.\n\n10. **Rodri (Manchester City)**\n    - Position: Midfielder\n    - Key Attributes: Defensive skills, passing, positional awareness.\n    - Achievements: Premier League titles, UEFA Champions League winner (2023).\n\nThis list is curated while adhering to the restriction of excluding Brazilian players. Each player included has demonstrated exceptional skills and remarkable performances, solidifying their status as some of the best in the world as of October 2023."}, {"role": "user", "content":
-      "You are not allowed to include Brazilian players"}], "model": "gpt-4o-mini", "stop": ["\nObservation:"]}'
-    headers:
-      accept:
-      - application/json
-      accept-encoding:
-      - gzip, deflate, zstd
-      connection:
-      - keep-alive
-      content-length:
-      - '9093'
-      content-type:
-      - application/json
-      cookie:
-      - __cf_bm=8Yv8F0ZCFAo2lf.qoqxao70yxyjVvIV90zQqVF6bVzQ-1749567342-1.0.1.1-fZgnv3RDfunvCO1koxwwFJrHnxSx_rwS_FHvQ6xxDPpKHwYr7dTqIQLZrNgSX5twGyK4F22rUmkuiS6KMVogcinChk8lmHtJBTUVTFjr2KU; _cfuvid=wzh8YnmXvLq1G0RcIVijtzboQtCZyIe2uZiochkBLqE-1749567342267-0.0.1.1-604800000
-      host:
-      - api.openai.com
-      user-agent:
-      - OpenAI/Python 1.78.0
-      x-stainless-arch:
-      - arm64
-      x-stainless-async:
-      - 'false'
-      x-stainless-lang:
-      - python
-      x-stainless-os:
-      - MacOS
-      x-stainless-package-version:
-      - 1.78.0
-      x-stainless-raw-response:
-      - 'true'
-      x-stainless-read-timeout:
-      - '600.0'
-      x-stainless-retry-count:
-      - '0'
-      x-stainless-runtime:
-      - CPython
-      x-stainless-runtime-version:
-      - 3.12.9
-    method: POST
-    uri: https://api.openai.com/v1/chat/completions
-  response:
-    body:
-      string: "{\n  \"id\": \"chatcmpl-BguhCefN1bN2OeYRo5ChhUqNBLUda\",\n  \"object\": \"chat.completion\",\n  \"created\": 1749567414,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n  \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\": \"assistant\",\n        \"content\": \"Thought: I now can give a great answer  \\nFinal Answer: Here is a refined list of the top 10 best soccer players in the world as of October 2023, ensuring that no Brazilian players are included:\\n\\n1. **Lionel Messi (Inter Miami CF)**\\n   - Position: Forward\\n   - Key Attributes: Dribbling, vision, goal-scoring ability.\\n   - Achievements: Multiple Ballon d'Or winner, Copa America champion, World Cup champion (2022).\\n\\n2. **Kylian Mbappé (Paris Saint-Germain)**\\n   - Position: Forward\\n   - Key Attributes: Speed, technique, finishing.\\n   - Achievements: FIFA World Cup champion (2018), multiple Ligue 1 titles, various domestic cups.\\n\\n3. **Erling Haaland (Manchester City)**\\\
-        n   - Position: Forward\\n   - Key Attributes: Power, speed, goal-scoring instinct.\\n   - Achievements: Bundesliga top scorer, UEFA Champions League winner (2023), Premier League titles.\\n\\n4. **Kevin De Bruyne (Manchester City)**\\n   - Position: Midfielder\\n   - Key Attributes: Passing, vision, creativity.\\n   - Achievements: Multiple Premier League titles, FA Cups, UEFA Champions League winner (2023).\\n\\n5. **Karim Benzema (Al-Ittihad)**\\n   - Position: Forward\\n   - Key Attributes: Goal-scoring, playmaking, tactical intelligence.\\n   - Achievements: 2022 Ballon d'Or winner, multiple Champions Leagues with Real Madrid.\\n\\n6. **Robert Lewandowski (FC Barcelona)**\\n   - Position: Forward\\n   - Key Attributes: Finishing, positioning, aerial ability.\\n   - Achievements: FIFA Best Men's Player, multiple Bundesliga titles, La Liga champion (2023).\\n\\n7. **Mohamed Salah (Liverpool)**\\n   - Position: Forward\\n   - Key Attributes: Speed, finishing, dribbling.\\n   -\
-        \ Achievements: Premier League champion, FA Cup, UEFA Champions League winner.\\n\\n8. **Luka Modrić (Real Madrid)**\\n   - Position: Midfielder\\n   - Key Attributes: Passing, vision, tactical intelligence.\\n   - Achievements: Multiple Champions League titles, Ballon d'Or winner (2018).\\n\\n9. **Harry Kane (Bayern Munich)**\\n   - Position: Forward\\n   - Key Attributes: Goal-scoring, technique, playmaking.\\n   - Achievements: Golden Boot winner, multiple Premier League titles, UEFA European Championship runner-up.\\n\\n10. **Son Heung-min (Tottenham Hotspur)**\\n    - Position: Forward\\n    - Key Attributes: Speed, finishing, playmaking.\\n    - Achievements: Premier League Golden Boot winner, multiple domestic cup titles.\\n\\nThis list has been carefully revised to exclude all Brazilian players while highlighting some of the most talented individuals in soccer as of October 2023. Each player has showcased remarkable effectiveness and skill, contributing significantly to their\
-        \ teams on both domestic and international stages.\",\n        \"refusal\": null,\n        \"annotations\": []\n      },\n      \"logprobs\": null,\n      \"finish_reason\": \"stop\"\n    }\n  ],\n  \"usage\": {\n    \"prompt_tokens\": 2028,\n    \"completion_tokens\": 614,\n    \"total_tokens\": 2642,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\": 1280,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\": {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\": 0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\": \"default\",\n  \"system_fingerprint\": \"fp_34a54ae93c\"\n}\n"
-    headers:
-      CF-RAY:
-      - 94d9b7d24d991d2c-GRU
-      Connection:
-      - keep-alive
-      Content-Type:
-      - application/json
-      Date:
-      - Tue, 10 Jun 2025 14:57:29 GMT
-      Server:
-      - cloudflare
-      Transfer-Encoding:
-      - chunked
-      X-Content-Type-Options:
-      - nosniff
-      access-control-expose-headers:
-      - X-Request-ID
-      alt-svc:
-      - h3=":443"; ma=86400
-      cf-cache-status:
-      - DYNAMIC
-      openai-organization:
-      - crewai-iuxna1
-      openai-processing-ms:
-      - '35291'
-      openai-version:
-      - '2020-10-01'
-      strict-transport-security:
-      - max-age=31536000; includeSubDomains; preload
-      x-envoy-upstream-service-time:
-      - '35294'
-      x-ratelimit-limit-requests:
-      - '30000'
-      x-ratelimit-limit-tokens:
-      - '150000000'
-      x-ratelimit-remaining-requests:
-      - '29999'
-      x-ratelimit-remaining-tokens:
-      - '149997855'
-      x-ratelimit-reset-requests:
-      - 2ms
-      x-ratelimit-reset-tokens:
-      - 0s
-      x-request-id:
-      - req_4676152d4227ac1825d1240ddef231d6
+      - X-REQUEST-ID-XXX
     status:
       code: 200
       message: OK

lib/crewai/tests/cassettes/agents/test_lite_agent_inside_flow_sync.yaml

@@ -1,14 +1,8 @@
 interactions:
 - request:
     body: '{"messages":[{"role":"system","content":"You are Test Agent. A helpful
-      test assistant\nYour personal goal is: Answer questions\nTo give my best complete
-      final answer to the task respond using the exact following format:\n\nThought:
-      I now can give a great answer\nFinal Answer: Your final answer must be the great
-      and the most complete as possible, it must be outcome described.\n\nI MUST use
-      these formats, my job depends on it!"},{"role":"user","content":"\nCurrent Task:
-      What is 2+2? Reply with just the number.\n\nBegin! This is VERY important to
-      you, use the tools available and give your best Final Answer, your job depends
-      on it!\n\nThought:"}],"model":"gpt-4o-mini"}'
+      test assistant\nYour personal goal is: Answer questions"},{"role":"user","content":"\nCurrent
+      Task: What is 2+2? Reply with just the number.\n\nProvide your complete response:"}],"model":"gpt-4o-mini"}'
     headers:
       User-Agent:
       - X-USER-AGENT-XXX
@@ -21,7 +15,7 @@ interactions:
       connection:
       - keep-alive
       content-length:
-      - '673'
+      - '272'
       content-type:
       - application/json
       host:
@@ -43,23 +37,22 @@ interactions:
       x-stainless-runtime:
       - CPython
       x-stainless-runtime-version:
-      - 3.13.3
+      - 3.13.5
     method: POST
     uri: https://api.openai.com/v1/chat/completions
   response:
     body:
-      string: "{\n  \"id\": \"chatcmpl-Cy7b0HjL79y39EkUcMLrRhPFe3XGj\",\n  \"object\":
-        \"chat.completion\",\n  \"created\": 1768444914,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n
+      string: "{\n  \"id\": \"chatcmpl-D6L4AzMHXLXDfyclWS6fJSwS0cvOl\",\n  \"object\":
+        \"chat.completion\",\n  \"created\": 1770403318,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n
         \ \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\":
-        \"assistant\",\n        \"content\": \"I now can give a great answer  \\nFinal
-        Answer: 4\",\n        \"refusal\": null,\n        \"annotations\": []\n      },\n
-        \     \"logprobs\": null,\n      \"finish_reason\": \"stop\"\n    }\n  ],\n
-        \ \"usage\": {\n    \"prompt_tokens\": 136,\n    \"completion_tokens\": 13,\n
-        \   \"total_tokens\": 149,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\":
+        \"assistant\",\n        \"content\": \"4\",\n        \"refusal\": null,\n
+        \       \"annotations\": []\n      },\n      \"logprobs\": null,\n      \"finish_reason\":
+        \"stop\"\n    }\n  ],\n  \"usage\": {\n    \"prompt_tokens\": 50,\n    \"completion_tokens\":
+        1,\n    \"total_tokens\": 51,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\":
         0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\":
         {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\":
         0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\":
-        \"default\",\n  \"system_fingerprint\": \"fp_8bbc38b4db\"\n}\n"
+        \"default\",\n  \"system_fingerprint\": \"fp_f4ae844694\"\n}\n"
     headers:
       CF-RAY:
       - CF-RAY-XXX
@@ -68,7 +61,7 @@ interactions:
       Content-Type:
       - application/json
       Date:
-      - Thu, 15 Jan 2026 02:41:55 GMT
+      - Fri, 06 Feb 2026 18:41:58 GMT
       Server:
       - cloudflare
       Set-Cookie:
@@ -85,18 +78,14 @@ interactions:
       - h3=":443"; ma=86400
       cf-cache-status:
       - DYNAMIC
-      content-length:
-      - '857'
       openai-organization:
       - OPENAI-ORG-XXX
       openai-processing-ms:
-      - '341'
+      - '264'
       openai-project:
       - OPENAI-PROJECT-XXX
       openai-version:
       - '2020-10-01'
-      x-envoy-upstream-service-time:
-      - '358'
       x-openai-proxy-wasm:
       - v0.1
       x-ratelimit-limit-requests:

lib/crewai/tests/cassettes/agents/test_lite_agent_standalone_still_works.yaml

@@ -1,14 +1,8 @@
 interactions:
 - request:
     body: '{"messages":[{"role":"system","content":"You are Standalone Agent. A helpful
-      assistant\nYour personal goal is: Answer questions\nTo give my best complete
-      final answer to the task respond using the exact following format:\n\nThought:
-      I now can give a great answer\nFinal Answer: Your final answer must be the great
-      and the most complete as possible, it must be outcome described.\n\nI MUST use
-      these formats, my job depends on it!"},{"role":"user","content":"\nCurrent Task:
-      What is 5+5? Reply with just the number.\n\nBegin! This is VERY important to
-      you, use the tools available and give your best Final Answer, your job depends
-      on it!\n\nThought:"}],"model":"gpt-4o-mini"}'
+      assistant\nYour personal goal is: Answer questions"},{"role":"user","content":"\nCurrent
+      Task: What is 5+5? Reply with just the number.\n\nProvide your complete response:"}],"model":"gpt-4o-mini"}'
     headers:
       User-Agent:
       - X-USER-AGENT-XXX
@@ -21,7 +15,7 @@ interactions:
       connection:
       - keep-alive
       content-length:
-      - '674'
+      - '273'
       content-type:
       - application/json
       host:
@@ -43,23 +37,22 @@ interactions:
       x-stainless-runtime:
       - CPython
       x-stainless-runtime-version:
-      - 3.13.3
+      - 3.13.5
     method: POST
     uri: https://api.openai.com/v1/chat/completions
   response:
     body:
-      string: "{\n  \"id\": \"chatcmpl-Cy7azhPwUHQ0p5tdhxSAmLPoE8UgC\",\n  \"object\":
-        \"chat.completion\",\n  \"created\": 1768444913,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n
+      string: "{\n  \"id\": \"chatcmpl-D6L3cLs2ndBaXV2wnqYCdi6X1ykvv\",\n  \"object\":
+        \"chat.completion\",\n  \"created\": 1770403284,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n
         \ \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\":
-        \"assistant\",\n        \"content\": \"I now can give a great answer  \\nFinal
-        Answer: 10\",\n        \"refusal\": null,\n        \"annotations\": []\n      },\n
-        \     \"logprobs\": null,\n      \"finish_reason\": \"stop\"\n    }\n  ],\n
-        \ \"usage\": {\n    \"prompt_tokens\": 136,\n    \"completion_tokens\": 13,\n
-        \   \"total_tokens\": 149,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\":
+        \"assistant\",\n        \"content\": \"10\",\n        \"refusal\": null,\n
+        \       \"annotations\": []\n      },\n      \"logprobs\": null,\n      \"finish_reason\":
+        \"stop\"\n    }\n  ],\n  \"usage\": {\n    \"prompt_tokens\": 50,\n    \"completion_tokens\":
+        1,\n    \"total_tokens\": 51,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\":
         0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\":
         {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\":
         0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\":
-        \"default\",\n  \"system_fingerprint\": \"fp_29330a9688\"\n}\n"
+        \"default\",\n  \"system_fingerprint\": \"fp_f4ae844694\"\n}\n"
     headers:
       CF-RAY:
       - CF-RAY-XXX
@@ -68,7 +61,7 @@ interactions:
       Content-Type:
       - application/json
       Date:
-      - Thu, 15 Jan 2026 02:41:54 GMT
+      - Fri, 06 Feb 2026 18:41:25 GMT
       Server:
       - cloudflare
       Set-Cookie:
@@ -85,18 +78,14 @@ interactions:
       - h3=":443"; ma=86400
       cf-cache-status:
       - DYNAMIC
-      content-length:
-      - '858'
       openai-organization:
       - OPENAI-ORG-XXX
       openai-processing-ms:
-      - '455'
+      - '270'
       openai-project:
       - OPENAI-PROJECT-XXX
       openai-version:
       - '2020-10-01'
-      x-envoy-upstream-service-time:
-      - '583'
       x-openai-proxy-wasm:
       - v0.1
       x-ratelimit-limit-requests:

lib/crewai/tests/cassettes/agents/test_multiple_agents_in_same_flow.yaml

@@ -1,13 +1,8 @@
 interactions:
 - request:
     body: '{"messages":[{"role":"system","content":"You are First Agent. A friendly
-      greeter\nYour personal goal is: Greet users\nTo give my best complete final
-      answer to the task respond using the exact following format:\n\nThought: I now
-      can give a great answer\nFinal Answer: Your final answer must be the great and
-      the most complete as possible, it must be outcome described.\n\nI MUST use these
-      formats, my job depends on it!"},{"role":"user","content":"\nCurrent Task: Say
-      hello\n\nBegin! This is VERY important to you, use the tools available and give
-      your best Final Answer, your job depends on it!\n\nThought:"}],"model":"gpt-4o-mini"}'
+      greeter\nYour personal goal is: Greet users"},{"role":"user","content":"\nCurrent
+      Task: Say hello\n\nProvide your complete response:"}],"model":"gpt-4o-mini"}'
     headers:
       User-Agent:
       - X-USER-AGENT-XXX
@@ -20,7 +15,7 @@ interactions:
       connection:
       - keep-alive
       content-length:
-      - '632'
+      - '231'
       content-type:
       - application/json
       host:
@@ -42,24 +37,22 @@ interactions:
       x-stainless-runtime:
       - CPython
       x-stainless-runtime-version:
-      - 3.13.3
+      - 3.13.5
     method: POST
     uri: https://api.openai.com/v1/chat/completions
   response:
     body:
-      string: "{\n  \"id\": \"chatcmpl-CyRKzgODZ9yn3F9OkaXsscLk2Ln3N\",\n  \"object\":
-        \"chat.completion\",\n  \"created\": 1768520801,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n
+      string: "{\n  \"id\": \"chatcmpl-D6L4A8Aad6P1YUxWjQpvyltn8GaKT\",\n  \"object\":
+        \"chat.completion\",\n  \"created\": 1770403318,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n
         \ \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\":
-        \"assistant\",\n        \"content\": \"I now can give a great answer  \\nFinal
-        Answer: Hello! Welcome! I'm so glad to see you here. If you need any assistance
-        or have any questions, feel free to ask. Have a wonderful day!\",\n        \"refusal\":
-        null,\n        \"annotations\": []\n      },\n      \"logprobs\": null,\n
-        \     \"finish_reason\": \"stop\"\n    }\n  ],\n  \"usage\": {\n    \"prompt_tokens\":
-        127,\n    \"completion_tokens\": 43,\n    \"total_tokens\": 170,\n    \"prompt_tokens_details\":
+        \"assistant\",\n        \"content\": \"Hello! \U0001F60A How are you today?\",\n
+        \       \"refusal\": null,\n        \"annotations\": []\n      },\n      \"logprobs\":
+        null,\n      \"finish_reason\": \"stop\"\n    }\n  ],\n  \"usage\": {\n    \"prompt_tokens\":
+        41,\n    \"completion_tokens\": 8,\n    \"total_tokens\": 49,\n    \"prompt_tokens_details\":
         {\n      \"cached_tokens\": 0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\":
         {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\":
         0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\":
-        \"default\",\n  \"system_fingerprint\": \"fp_c4585b5b9c\"\n}\n"
+        \"default\",\n  \"system_fingerprint\": \"fp_f4ae844694\"\n}\n"
     headers:
       CF-RAY:
       - CF-RAY-XXX
@@ -68,7 +61,7 @@ interactions:
       Content-Type:
       - application/json
       Date:
-      - Thu, 15 Jan 2026 23:46:42 GMT
+      - Fri, 06 Feb 2026 18:41:58 GMT
       Server:
       - cloudflare
       Set-Cookie:
@@ -85,18 +78,14 @@ interactions:
       - h3=":443"; ma=86400
       cf-cache-status:
       - DYNAMIC
-      content-length:
-      - '990'
       openai-organization:
       - OPENAI-ORG-XXX
       openai-processing-ms:
-      - '880'
+      - '325'
       openai-project:
       - OPENAI-PROJECT-XXX
       openai-version:
       - '2020-10-01'
-      x-envoy-upstream-service-time:
-      - '1160'
       x-openai-proxy-wasm:
       - v0.1
       x-ratelimit-limit-requests:
@@ -118,13 +107,8 @@ interactions:
       message: OK
 - request:
     body: '{"messages":[{"role":"system","content":"You are Second Agent. A polite
-      farewell agent\nYour personal goal is: Say goodbye\nTo give my best complete
-      final answer to the task respond using the exact following format:\n\nThought:
-      I now can give a great answer\nFinal Answer: Your final answer must be the great
-      and the most complete as possible, it must be outcome described.\n\nI MUST use
-      these formats, my job depends on it!"},{"role":"user","content":"\nCurrent Task:
-      Say goodbye\n\nBegin! This is VERY important to you, use the tools available
-      and give your best Final Answer, your job depends on it!\n\nThought:"}],"model":"gpt-4o-mini"}'
+      farewell agent\nYour personal goal is: Say goodbye"},{"role":"user","content":"\nCurrent
+      Task: Say goodbye\n\nProvide your complete response:"}],"model":"gpt-4o-mini"}'
     headers:
       User-Agent:
       - X-USER-AGENT-XXX
@@ -137,7 +121,7 @@ interactions:
       connection:
       - keep-alive
       content-length:
-      - '640'
+      - '239'
       content-type:
       - application/json
       host:
@@ -159,27 +143,24 @@ interactions:
       x-stainless-runtime:
       - CPython
       x-stainless-runtime-version:
-      - 3.13.3
+      - 3.13.5
     method: POST
     uri: https://api.openai.com/v1/chat/completions
   response:
     body:
-      string: "{\n  \"id\": \"chatcmpl-CyRL1Ua2PkK5xXPp3KeF0AnGAk3JP\",\n  \"object\":
-        \"chat.completion\",\n  \"created\": 1768520803,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n
+      string: "{\n  \"id\": \"chatcmpl-D6L4BLMYC3ODccwbKfBIdtrEyd3no\",\n  \"object\":
+        \"chat.completion\",\n  \"created\": 1770403319,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n
         \ \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\":
-        \"assistant\",\n        \"content\": \"I now can give a great answer  \\nFinal
-        Answer: As we reach the end of our conversation, I want to express my gratitude
-        for the time we've shared. It's been a pleasure assisting you, and I hope
-        you found our interaction helpful and enjoyable. Remember, whenever you need
-        assistance, I'm just a message away. Wishing you all the best in your future
-        endeavors. Goodbye and take care!\",\n        \"refusal\": null,\n        \"annotations\":
+        \"assistant\",\n        \"content\": \"Thank you for the time we've spent
+        together! I wish you all the best in your future endeavors. Take care, and
+        until we meet again, goodbye!\",\n        \"refusal\": null,\n        \"annotations\":
         []\n      },\n      \"logprobs\": null,\n      \"finish_reason\": \"stop\"\n
-        \   }\n  ],\n  \"usage\": {\n    \"prompt_tokens\": 126,\n    \"completion_tokens\":
-        79,\n    \"total_tokens\": 205,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\":
+        \   }\n  ],\n  \"usage\": {\n    \"prompt_tokens\": 40,\n    \"completion_tokens\":
+        31,\n    \"total_tokens\": 71,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\":
         0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\":
         {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\":
         0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\":
-        \"default\",\n  \"system_fingerprint\": \"fp_29330a9688\"\n}\n"
+        \"default\",\n  \"system_fingerprint\": \"fp_f4ae844694\"\n}\n"
     headers:
       CF-RAY:
       - CF-RAY-XXX
@@ -188,7 +169,7 @@ interactions:
       Content-Type:
       - application/json
       Date:
-      - Thu, 15 Jan 2026 23:46:44 GMT
+      - Fri, 06 Feb 2026 18:41:59 GMT
       Server:
       - cloudflare
       Set-Cookie:
@@ -205,18 +186,14 @@ interactions:
       - h3=":443"; ma=86400
       cf-cache-status:
       - DYNAMIC
-      content-length:
-      - '1189'
       openai-organization:
       - OPENAI-ORG-XXX
       openai-processing-ms:
-      - '1363'
+      - '726'
       openai-project:
       - OPENAI-PROJECT-XXX
       openai-version:
       - '2020-10-01'
-      x-envoy-upstream-service-time:
-      - '1605'
       x-openai-proxy-wasm:
       - v0.1
       x-ratelimit-limit-requests:

lib/crewai/tests/cassettes/agents/test_native_tool_calling_error_handling.yaml

@@ -2,9 +2,8 @@ interactions:
 - request:
     body: '{"messages":[{"role":"system","content":"You are Calculator. You calculate
       things.\nYour personal goal is: Perform calculations efficiently"},{"role":"user","content":"\nCurrent
-      Task: Use the failing_tool to do something.\n\nThis is VERY important to you,
-      your job depends on it!"}],"model":"gpt-4o-mini","tool_choice":"auto","tools":[{"type":"function","function":{"name":"failing_tool","description":"This
-      tool always fails","parameters":{"properties":{},"type":"object"}}}]}'
+      Task: Use the failing_tool to do something."}],"model":"gpt-4o-mini","tool_choice":"auto","tools":[{"type":"function","function":{"name":"failing_tool","description":"This
+      tool always fails","strict":true,"parameters":{"properties":{},"type":"object","additionalProperties":false,"required":[]}}}]}'
     headers:
       User-Agent:
       - X-USER-AGENT-XXX
@@ -17,7 +16,7 @@ interactions:
       connection:
       - keep-alive
       content-length:
-      - '477'
+      - '476'
       content-type:
       - application/json
       host:
@@ -39,26 +38,26 @@ interactions:
       x-stainless-runtime:
       - CPython
       x-stainless-runtime-version:
-      - 3.13.3
+      - 3.13.5
     method: POST
     uri: https://api.openai.com/v1/chat/completions
   response:
     body:
-      string: "{\n  \"id\": \"chatcmpl-D0vm2JDsOmy0czXPAr4vnw3wvuqYZ\",\n  \"object\":
-        \"chat.completion\",\n  \"created\": 1769114454,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n
+      string: "{\n  \"id\": \"chatcmpl-D6L3dV6acwapgRyxmnzGfuOXemtjJ\",\n  \"object\":
+        \"chat.completion\",\n  \"created\": 1770403285,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n
         \ \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\":
         \"assistant\",\n        \"content\": null,\n        \"tool_calls\": [\n          {\n
-        \           \"id\": \"call_8xr8rPUDWzLfQ3LOWPHtBUjK\",\n            \"type\":
+        \           \"id\": \"call_GCdaOdo32pr1sSk4RzO0tiB9\",\n            \"type\":
         \"function\",\n            \"function\": {\n              \"name\": \"failing_tool\",\n
         \             \"arguments\": \"{}\"\n            }\n          }\n        ],\n
         \       \"refusal\": null,\n        \"annotations\": []\n      },\n      \"logprobs\":
         null,\n      \"finish_reason\": \"tool_calls\"\n    }\n  ],\n  \"usage\":
-        {\n    \"prompt_tokens\": 78,\n    \"completion_tokens\": 11,\n    \"total_tokens\":
-        89,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\": 0,\n      \"audio_tokens\":
+        {\n    \"prompt_tokens\": 65,\n    \"completion_tokens\": 11,\n    \"total_tokens\":
+        76,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\": 0,\n      \"audio_tokens\":
         0\n    },\n    \"completion_tokens_details\": {\n      \"reasoning_tokens\":
         0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\": 0,\n
         \     \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\":
-        \"default\",\n  \"system_fingerprint\": \"fp_c4585b5b9c\"\n}\n"
+        \"default\",\n  \"system_fingerprint\": \"fp_6c0d1490cb\"\n}\n"
     headers:
       CF-RAY:
       - CF-RAY-XXX
@@ -67,7 +66,7 @@ interactions:
       Content-Type:
       - application/json
       Date:
-      - Thu, 22 Jan 2026 20:40:54 GMT
+      - Fri, 06 Feb 2026 18:41:25 GMT
       Server:
       - cloudflare
       Set-Cookie:
@@ -87,13 +86,11 @@ interactions:
       openai-organization:
       - OPENAI-ORG-XXX
       openai-processing-ms:
-      - '593'
+      - '436'
       openai-project:
       - OPENAI-PROJECT-XXX
       openai-version:
       - '2020-10-01'
-      x-envoy-upstream-service-time:
-      - '621'
       x-openai-proxy-wasm:
       - v0.1
       x-ratelimit-limit-requests:
@@ -116,12 +113,9 @@ interactions:
 - request:
     body: '{"messages":[{"role":"system","content":"You are Calculator. You calculate
       things.\nYour personal goal is: Perform calculations efficiently"},{"role":"user","content":"\nCurrent
-      Task: Use the failing_tool to do something.\n\nThis is VERY important to you,
-      your job depends on it!"},{"role":"assistant","content":null,"tool_calls":[{"id":"call_8xr8rPUDWzLfQ3LOWPHtBUjK","type":"function","function":{"name":"failing_tool","arguments":"{}"}}]},{"role":"tool","tool_call_id":"call_8xr8rPUDWzLfQ3LOWPHtBUjK","content":"Error
-      executing tool: This tool always fails"},{"role":"user","content":"Analyze the
-      tool result. If requirements are met, provide the Final Answer. Otherwise, call
-      the next tool. Deliver only the answer without meta-commentary."}],"model":"gpt-4o-mini","tool_choice":"auto","tools":[{"type":"function","function":{"name":"failing_tool","description":"This
-      tool always fails","parameters":{"properties":{},"type":"object"}}}]}'
+      Task: Use the failing_tool to do something."},{"role":"assistant","content":null,"tool_calls":[{"id":"call_GCdaOdo32pr1sSk4RzO0tiB9","type":"function","function":{"name":"failing_tool","arguments":"{}"}}]},{"role":"tool","tool_call_id":"call_GCdaOdo32pr1sSk4RzO0tiB9","name":"failing_tool","content":"Error
+      executing tool: This tool always fails"}],"model":"gpt-4o-mini","tool_choice":"auto","tools":[{"type":"function","function":{"name":"failing_tool","description":"This
+      tool always fails","strict":true,"parameters":{"properties":{},"type":"object","additionalProperties":false,"required":[]}}}]}'
     headers:
       User-Agent:
       - X-USER-AGENT-XXX
@@ -134,7 +128,7 @@ interactions:
       connection:
       - keep-alive
       content-length:
-      - '941'
+      - '778'
       content-type:
       - application/json
       cookie:
@@ -158,22 +152,25 @@ interactions:
       x-stainless-runtime:
       - CPython
       x-stainless-runtime-version:
-      - 3.13.3
+      - 3.13.5
     method: POST
     uri: https://api.openai.com/v1/chat/completions
   response:
     body:
-      string: "{\n  \"id\": \"chatcmpl-D0vm3xcywoKBW75bhBXfkGJNim6Th\",\n  \"object\":
-        \"chat.completion\",\n  \"created\": 1769114455,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n
+      string: "{\n  \"id\": \"chatcmpl-D6L3dhjDZOoihHvXvRpbJD3ReGu0z\",\n  \"object\":
+        \"chat.completion\",\n  \"created\": 1770403285,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n
         \ \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\":
-        \"assistant\",\n        \"content\": \"Error: This tool always fails.\",\n
-        \       \"refusal\": null,\n        \"annotations\": []\n      },\n      \"logprobs\":
-        null,\n      \"finish_reason\": \"stop\"\n    }\n  ],\n  \"usage\": {\n    \"prompt_tokens\":
-        141,\n    \"completion_tokens\": 8,\n    \"total_tokens\": 149,\n    \"prompt_tokens_details\":
-        {\n      \"cached_tokens\": 0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\":
+        \"assistant\",\n        \"content\": \"The attempt to use the failing tool
+        resulted in an error, as expected since it is designed to always fail. If
+        there's anything else you would like to calculate or explore, please let me
+        know!\",\n        \"refusal\": null,\n        \"annotations\": []\n      },\n
+        \     \"logprobs\": null,\n      \"finish_reason\": \"stop\"\n    }\n  ],\n
+        \ \"usage\": {\n    \"prompt_tokens\": 93,\n    \"completion_tokens\": 40,\n
+        \   \"total_tokens\": 133,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\":
+        0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\":
         {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\":
         0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\":
-        \"default\",\n  \"system_fingerprint\": \"fp_c4585b5b9c\"\n}\n"
+        \"default\",\n  \"system_fingerprint\": \"fp_6c0d1490cb\"\n}\n"
     headers:
       CF-RAY:
       - CF-RAY-XXX
@@ -182,7 +179,7 @@ interactions:
       Content-Type:
       - application/json
       Date:
-      - Thu, 22 Jan 2026 20:40:55 GMT
+      - Fri, 06 Feb 2026 18:41:26 GMT
       Server:
       - cloudflare
       Strict-Transport-Security:
@@ -200,13 +197,11 @@ interactions:
       openai-organization:
       - OPENAI-ORG-XXX
       openai-processing-ms:
-      - '420'
+      - '776'
       openai-project:
       - OPENAI-PROJECT-XXX
       openai-version:
       - '2020-10-01'
-      x-envoy-upstream-service-time:
-      - '436'
       x-openai-proxy-wasm:
       - v0.1
       x-ratelimit-limit-requests:

lib/crewai/tests/cassettes/llms/anthropic/test_anthropic_agent_kickoff_structured_output_with_tools.yaml

@@ -43,15 +43,15 @@ interactions:
       x-stainless-runtime:
       - CPython
       x-stainless-runtime-version:
-      - 3.13.3
+      - 3.13.5
       x-stainless-timeout:
       - NOT_GIVEN
     method: POST
     uri: https://api.anthropic.com/v1/messages
   response:
     body:
-      string: '{"model":"claude-3-5-haiku-20241022","id":"msg_0149zKBgM47utdBdrfJjM6YZ","type":"message","role":"assistant","content":[{"type":"tool_use","id":"toolu_011jnBYLgtzXqdmSi7JDyQHj","name":"structured_output","input":{"operation":"Addition","result":42,"explanation":"Adding
-        15 and 27 together results in 42"}}],"stop_reason":"tool_use","stop_sequence":null,"usage":{"input_tokens":573,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"cache_creation":{"ephemeral_5m_input_tokens":0,"ephemeral_1h_input_tokens":0},"output_tokens":79,"service_tier":"standard"}}'
+      string: '{"model":"claude-3-5-haiku-20241022","id":"msg_01A41GpDoJbZLUhR8dQzUcUX","type":"message","role":"assistant","content":[{"type":"tool_use","id":"toolu_01UNPdzpayoWyqDYVE7fR5oA","name":"structured_output","input":{"operation":"Addition","result":42,"explanation":"Added
+        15 and 27 together"}}],"stop_reason":"tool_use","stop_sequence":null,"usage":{"input_tokens":573,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"cache_creation":{"ephemeral_5m_input_tokens":0,"ephemeral_1h_input_tokens":0},"output_tokens":75,"service_tier":"standard","inference_geo":"not_available"}}'
     headers:
       CF-RAY:
       - CF-RAY-XXX
@@ -62,7 +62,7 @@ interactions:
       Content-Type:
       - application/json
       Date:
-      - Fri, 30 Jan 2026 18:56:15 GMT
+      - Fri, 06 Feb 2026 18:41:25 GMT
       Server:
       - cloudflare
       Transfer-Encoding:
@@ -88,7 +88,7 @@ interactions:
       anthropic-ratelimit-requests-remaining:
       - '3999'
       anthropic-ratelimit-requests-reset:
-      - '2026-01-30T18:56:14Z'
+      - '2026-02-06T18:41:24Z'
       anthropic-ratelimit-tokens-limit:
       - ANTHROPIC-RATELIMIT-TOKENS-LIMIT-XXX
       anthropic-ratelimit-tokens-remaining:
@@ -102,7 +102,7 @@ interactions:
       strict-transport-security:
       - STS-XXX
       x-envoy-upstream-service-time:
-      - '1473'
+      - '1247'
     status:
       code: 200
       message: OK

lib/crewai/tests/cassettes/llms/anthropic/test_anthropic_agent_kickoff_structured_output_without_tools.yaml

@@ -44,21 +44,20 @@ interactions:
       x-stainless-runtime:
       - CPython
       x-stainless-runtime-version:
-      - 3.13.3
+      - 3.13.5
       x-stainless-timeout:
       - NOT_GIVEN
     method: POST
     uri: https://api.anthropic.com/v1/messages
   response:
     body:
-      string: '{"model":"claude-3-5-haiku-20241022","id":"msg_013iHkpmto99iyH5kDvn8uER","type":"message","role":"assistant","content":[{"type":"tool_use","id":"toolu_01Kpda2DzHBqWq9a2FS2Bdw6","name":"structured_output","input":{"topic":"Benefits
+      string: '{"model":"claude-3-5-haiku-20241022","id":"msg_016wrV83wm3FLYD4JoTy2Piw","type":"message","role":"assistant","content":[{"type":"tool_use","id":"toolu_01V6Pzr7eGfuG4Q3mc25ZXwN","name":"structured_output","input":{"topic":"Benefits
         of Remote Work","summary":"Remote work offers significant advantages for both
-        employees and employers, transforming traditional work paradigms by providing
-        flexibility, increased productivity, and cost savings.","key_points":["Increased
-        employee flexibility and work-life balance","Reduced commuting time and associated
-        stress","Cost savings for companies on office infrastructure","Access to a
-        global talent pool","Higher employee productivity and job satisfaction","Lower
-        carbon footprint due to reduced travel"]}}],"stop_reason":"tool_use","stop_sequence":null,"usage":{"input_tokens":589,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"cache_creation":{"ephemeral_5m_input_tokens":0,"ephemeral_1h_input_tokens":0},"output_tokens":153,"service_tier":"standard"}}'
+        employees and employers, transforming traditional workplace dynamics.","key_points":["Increased
+        flexibility in work schedule","Reduced commute time and transportation costs","Improved
+        work-life balance","Higher productivity for many employees","Cost savings
+        for companies on office infrastructure","Expanded talent pool for hiring","Enhanced
+        employee job satisfaction"]}}],"stop_reason":"tool_use","stop_sequence":null,"usage":{"input_tokens":589,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"cache_creation":{"ephemeral_5m_input_tokens":0,"ephemeral_1h_input_tokens":0},"output_tokens":142,"service_tier":"standard","inference_geo":"not_available"}}'
     headers:
       CF-RAY:
       - CF-RAY-XXX
@@ -69,7 +68,7 @@ interactions:
       Content-Type:
       - application/json
       Date:
-      - Fri, 30 Jan 2026 18:56:19 GMT
+      - Fri, 06 Feb 2026 18:41:28 GMT
       Server:
       - cloudflare
       Transfer-Encoding:
@@ -95,7 +94,7 @@ interactions:
       anthropic-ratelimit-requests-remaining:
       - '3999'
       anthropic-ratelimit-requests-reset:
-      - '2026-01-30T18:56:16Z'
+      - '2026-02-06T18:41:26Z'
       anthropic-ratelimit-tokens-limit:
       - ANTHROPIC-RATELIMIT-TOKENS-LIMIT-XXX
       anthropic-ratelimit-tokens-remaining:
@@ -109,7 +108,7 @@ interactions:
       strict-transport-security:
       - STS-XXX
       x-envoy-upstream-service-time:
-      - '3107'
+      - '2650'
     status:
       code: 200
       message: OK

lib/crewai/tests/cassettes/llms/anthropic/test_anthropic_tool_execution_returns_tool_result_directly.yaml

@@ -0,0 +1,102 @@
+interactions:
+- request:
+    body: '{"max_tokens":4096,"messages":[{"role":"user","content":"Calculate 5 +
+      3 using the simple_calculator tool with operation ''add''."}],"model":"claude-3-5-haiku-20241022","stream":false,"tool_choice":{"type":"tool","name":"simple_calculator"},"tools":[{"name":"simple_calculator","description":"Perform
+      simple math operations","input_schema":{"type":"object","properties":{"operation":{"type":"string","enum":["add","multiply"],"description":"The
+      operation to perform"},"a":{"type":"integer","description":"First number"},"b":{"type":"integer","description":"Second
+      number"}},"required":["operation","a","b"]}}]}'
+    headers:
+      accept:
+      - application/json
+      accept-encoding:
+      - ACCEPT-ENCODING-XXX
+      anthropic-version:
+      - '2023-06-01'
+      connection:
+      - keep-alive
+      content-length:
+      - '608'
+      content-type:
+      - application/json
+      host:
+      - api.anthropic.com
+      user-agent:
+      - X-USER-AGENT-XXX
+      x-stainless-arch:
+      - X-STAINLESS-ARCH-XXX
+      x-stainless-async:
+      - 'false'
+      x-stainless-lang:
+      - python
+      x-stainless-os:
+      - X-STAINLESS-OS-XXX
+      x-stainless-package-version:
+      - 0.73.0
+      x-stainless-retry-count:
+      - '0'
+      x-stainless-runtime:
+      - CPython
+      x-stainless-runtime-version:
+      - 3.13.3
+      x-stainless-timeout:
+      - NOT_GIVEN
+    method: POST
+    uri: https://api.anthropic.com/v1/messages
+  response:
+    body:
+      string: '{"model":"claude-3-5-haiku-20241022","id":"msg_01Q2F83aAeqqTCxsd8WpZjK7","type":"message","role":"assistant","content":[{"type":"tool_use","id":"toolu_01BW4XkHnhRVM5JZsvoaQKw5","name":"simple_calculator","input":{"operation":"add","a":5,"b":3}}],"stop_reason":"tool_use","stop_sequence":null,"usage":{"input_tokens":498,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"cache_creation":{"ephemeral_5m_input_tokens":0,"ephemeral_1h_input_tokens":0},"output_tokens":67,"service_tier":"standard"}}'
+    headers:
+      CF-RAY:
+      - CF-RAY-XXX
+      Connection:
+      - keep-alive
+      Content-Security-Policy:
+      - CSP-FILTERED
+      Content-Type:
+      - application/json
+      Date:
+      - Tue, 03 Feb 2026 23:26:35 GMT
+      Server:
+      - cloudflare
+      Transfer-Encoding:
+      - chunked
+      X-Robots-Tag:
+      - none
+      anthropic-organization-id:
+      - ANTHROPIC-ORGANIZATION-ID-XXX
+      anthropic-ratelimit-input-tokens-limit:
+      - ANTHROPIC-RATELIMIT-INPUT-TOKENS-LIMIT-XXX
+      anthropic-ratelimit-input-tokens-remaining:
+      - ANTHROPIC-RATELIMIT-INPUT-TOKENS-REMAINING-XXX
+      anthropic-ratelimit-input-tokens-reset:
+      - ANTHROPIC-RATELIMIT-INPUT-TOKENS-RESET-XXX
+      anthropic-ratelimit-output-tokens-limit:
+      - ANTHROPIC-RATELIMIT-OUTPUT-TOKENS-LIMIT-XXX
+      anthropic-ratelimit-output-tokens-remaining:
+      - ANTHROPIC-RATELIMIT-OUTPUT-TOKENS-REMAINING-XXX
+      anthropic-ratelimit-output-tokens-reset:
+      - ANTHROPIC-RATELIMIT-OUTPUT-TOKENS-RESET-XXX
+      anthropic-ratelimit-requests-limit:
+      - '4000'
+      anthropic-ratelimit-requests-remaining:
+      - '3999'
+      anthropic-ratelimit-requests-reset:
+      - '2026-02-03T23:26:34Z'
+      anthropic-ratelimit-tokens-limit:
+      - ANTHROPIC-RATELIMIT-TOKENS-LIMIT-XXX
+      anthropic-ratelimit-tokens-remaining:
+      - ANTHROPIC-RATELIMIT-TOKENS-REMAINING-XXX
+      anthropic-ratelimit-tokens-reset:
+      - ANTHROPIC-RATELIMIT-TOKENS-RESET-XXX
+      cf-cache-status:
+      - DYNAMIC
+      request-id:
+      - REQUEST-ID-XXX
+      strict-transport-security:
+      - STS-XXX
+      x-envoy-upstream-service-time:
+      - '1228'
+    status:
+      code: 200
+      message: OK
+version: 1

lib/crewai/tests/cassettes/llms/anthropic/test_anthropic_tool_execution_with_available_functions.yaml

@@ -0,0 +1,108 @@
+interactions:
+- request:
+    body: '{"max_tokens":4096,"messages":[{"role":"user","content":"Create a simple
+      plan to say hello. Use the create_reasoning_plan tool."}],"model":"claude-3-5-haiku-20241022","stream":false,"tool_choice":{"type":"tool","name":"create_reasoning_plan"},"tools":[{"name":"create_reasoning_plan","description":"Create
+      a structured reasoning plan for completing a task","input_schema":{"type":"object","properties":{"plan":{"type":"string","description":"High-level
+      plan description"},"steps":{"type":"array","items":{"type":"object"},"description":"List
+      of steps to execute"},"ready":{"type":"boolean","description":"Whether the plan
+      is ready to execute"}},"required":["plan","steps","ready"]}}]}'
+    headers:
+      accept:
+      - application/json
+      accept-encoding:
+      - ACCEPT-ENCODING-XXX
+      anthropic-version:
+      - '2023-06-01'
+      connection:
+      - keep-alive
+      content-length:
+      - '684'
+      content-type:
+      - application/json
+      host:
+      - api.anthropic.com
+      user-agent:
+      - X-USER-AGENT-XXX
+      x-stainless-arch:
+      - X-STAINLESS-ARCH-XXX
+      x-stainless-async:
+      - 'false'
+      x-stainless-lang:
+      - python
+      x-stainless-os:
+      - X-STAINLESS-OS-XXX
+      x-stainless-package-version:
+      - 0.73.0
+      x-stainless-retry-count:
+      - '0'
+      x-stainless-runtime:
+      - CPython
+      x-stainless-runtime-version:
+      - 3.13.3
+      x-stainless-timeout:
+      - NOT_GIVEN
+    method: POST
+    uri: https://api.anthropic.com/v1/messages
+  response:
+    body:
+      string: '{"model":"claude-3-5-haiku-20241022","id":"msg_01HLuGgGRFseMdhTYAhkKtfz","type":"message","role":"assistant","content":[{"type":"tool_use","id":"toolu_01GQAUFHffGzMd3ufA6YRMZF","name":"create_reasoning_plan","input":{"plan":"Say
+        hello in a friendly and straightforward manner","steps":[{"description":"Take
+        a deep breath","action":"Pause and relax"},{"description":"Smile","action":"Prepare
+        a warm facial expression"},{"description":"Greet the person","action":"Say
+        ''Hello!''"},{"description":"Wait for response","action":"Listen and be ready
+        to continue conversation"}],"ready":true}}],"stop_reason":"tool_use","stop_sequence":null,"usage":{"input_tokens":513,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"cache_creation":{"ephemeral_5m_input_tokens":0,"ephemeral_1h_input_tokens":0},"output_tokens":162,"service_tier":"standard"}}'
+    headers:
+      CF-RAY:
+      - CF-RAY-XXX
+      Connection:
+      - keep-alive
+      Content-Security-Policy:
+      - CSP-FILTERED
+      Content-Type:
+      - application/json
+      Date:
+      - Tue, 03 Feb 2026 23:26:38 GMT
+      Server:
+      - cloudflare
+      Transfer-Encoding:
+      - chunked
+      X-Robots-Tag:
+      - none
+      anthropic-organization-id:
+      - ANTHROPIC-ORGANIZATION-ID-XXX
+      anthropic-ratelimit-input-tokens-limit:
+      - ANTHROPIC-RATELIMIT-INPUT-TOKENS-LIMIT-XXX
+      anthropic-ratelimit-input-tokens-remaining:
+      - ANTHROPIC-RATELIMIT-INPUT-TOKENS-REMAINING-XXX
+      anthropic-ratelimit-input-tokens-reset:
+      - ANTHROPIC-RATELIMIT-INPUT-TOKENS-RESET-XXX
+      anthropic-ratelimit-output-tokens-limit:
+      - ANTHROPIC-RATELIMIT-OUTPUT-TOKENS-LIMIT-XXX
+      anthropic-ratelimit-output-tokens-remaining:
+      - ANTHROPIC-RATELIMIT-OUTPUT-TOKENS-REMAINING-XXX
+      anthropic-ratelimit-output-tokens-reset:
+      - ANTHROPIC-RATELIMIT-OUTPUT-TOKENS-RESET-XXX
+      anthropic-ratelimit-requests-limit:
+      - '4000'
+      anthropic-ratelimit-requests-remaining:
+      - '3999'
+      anthropic-ratelimit-requests-reset:
+      - '2026-02-03T23:26:35Z'
+      anthropic-ratelimit-tokens-limit:
+      - ANTHROPIC-RATELIMIT-TOKENS-LIMIT-XXX
+      anthropic-ratelimit-tokens-remaining:
+      - ANTHROPIC-RATELIMIT-TOKENS-REMAINING-XXX
+      anthropic-ratelimit-tokens-reset:
+      - ANTHROPIC-RATELIMIT-TOKENS-RESET-XXX
+      cf-cache-status:
+      - DYNAMIC
+      request-id:
+      - REQUEST-ID-XXX
+      strict-transport-security:
+      - STS-XXX
+      x-envoy-upstream-service-time:
+      - '2994'
+    status:
+      code: 200
+      message: OK
+version: 1

lib/crewai/tests/cassettes/utilities/test_llm_call_events_share_call_id.yaml

@@ -0,0 +1,108 @@
+interactions:
+- request:
+    body: '{"messages":[{"role":"user","content":"Say hi"}],"model":"gpt-4o-mini"}'
+    headers:
+      User-Agent:
+      - X-USER-AGENT-XXX
+      accept:
+      - application/json
+      accept-encoding:
+      - ACCEPT-ENCODING-XXX
+      authorization:
+      - AUTHORIZATION-XXX
+      connection:
+      - keep-alive
+      content-length:
+      - '71'
+      content-type:
+      - application/json
+      host:
+      - api.openai.com
+      x-stainless-arch:
+      - X-STAINLESS-ARCH-XXX
+      x-stainless-async:
+      - 'false'
+      x-stainless-lang:
+      - python
+      x-stainless-os:
+      - X-STAINLESS-OS-XXX
+      x-stainless-package-version:
+      - 1.83.0
+      x-stainless-read-timeout:
+      - X-STAINLESS-READ-TIMEOUT-XXX
+      x-stainless-retry-count:
+      - '0'
+      x-stainless-runtime:
+      - CPython
+      x-stainless-runtime-version:
+      - 3.13.0
+    method: POST
+    uri: https://api.openai.com/v1/chat/completions
+  response:
+    body:
+      string: "{\n  \"id\": \"chatcmpl-D2HpUSxS5LeHwDTELElWlC5CDMzmr\",\n  \"object\":
+        \"chat.completion\",\n  \"created\": 1769437564,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n
+        \ \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\":
+        \"assistant\",\n        \"content\": \"Hi there! How can I assist you today?\",\n
+        \       \"refusal\": null,\n        \"annotations\": []\n      },\n      \"logprobs\":
+        null,\n      \"finish_reason\": \"stop\"\n    }\n  ],\n  \"usage\": {\n    \"prompt_tokens\":
+        9,\n    \"completion_tokens\": 10,\n    \"total_tokens\": 19,\n    \"prompt_tokens_details\":
+        {\n      \"cached_tokens\": 0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\":
+        {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\":
+        0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\":
+        \"default\",\n  \"system_fingerprint\": \"fp_29330a9688\"\n}\n"
+    headers:
+      CF-RAY:
+      - CF-RAY-XXX
+      Connection:
+      - keep-alive
+      Content-Type:
+      - application/json
+      Date:
+      - Mon, 26 Jan 2026 14:26:05 GMT
+      Server:
+      - cloudflare
+      Set-Cookie:
+      - SET-COOKIE-XXX
+      Strict-Transport-Security:
+      - STS-XXX
+      Transfer-Encoding:
+      - chunked
+      X-Content-Type-Options:
+      - X-CONTENT-TYPE-XXX
+      access-control-expose-headers:
+      - ACCESS-CONTROL-XXX
+      alt-svc:
+      - h3=":443"; ma=86400
+      cf-cache-status:
+      - DYNAMIC
+      openai-organization:
+      - OPENAI-ORG-XXX
+      openai-processing-ms:
+      - '460'
+      openai-project:
+      - OPENAI-PROJECT-XXX
+      openai-version:
+      - '2020-10-01'
+      x-envoy-upstream-service-time:
+      - '477'
+      x-openai-proxy-wasm:
+      - v0.1
+      x-ratelimit-limit-requests:
+      - X-RATELIMIT-LIMIT-REQUESTS-XXX
+      x-ratelimit-limit-tokens:
+      - X-RATELIMIT-LIMIT-TOKENS-XXX
+      x-ratelimit-remaining-requests:
+      - X-RATELIMIT-REMAINING-REQUESTS-XXX
+      x-ratelimit-remaining-tokens:
+      - X-RATELIMIT-REMAINING-TOKENS-XXX
+      x-ratelimit-reset-requests:
+      - X-RATELIMIT-RESET-REQUESTS-XXX
+      x-ratelimit-reset-tokens:
+      - X-RATELIMIT-RESET-TOKENS-XXX
+      x-request-id:
+      - X-REQUEST-ID-XXX
+    status:
+      code: 200
+      message: OK
+version: 1

lib/crewai/tests/cassettes/utilities/test_separate_llm_calls_have_different_call_ids.yaml

@@ -0,0 +1,215 @@
+interactions:
+- request:
+    body: '{"messages":[{"role":"user","content":"Say hi"}],"model":"gpt-4o-mini"}'
+    headers:
+      User-Agent:
+      - X-USER-AGENT-XXX
+      accept:
+      - application/json
+      accept-encoding:
+      - ACCEPT-ENCODING-XXX
+      authorization:
+      - AUTHORIZATION-XXX
+      connection:
+      - keep-alive
+      content-length:
+      - '71'
+      content-type:
+      - application/json
+      host:
+      - api.openai.com
+      x-stainless-arch:
+      - X-STAINLESS-ARCH-XXX
+      x-stainless-async:
+      - 'false'
+      x-stainless-lang:
+      - python
+      x-stainless-os:
+      - X-STAINLESS-OS-XXX
+      x-stainless-package-version:
+      - 1.83.0
+      x-stainless-read-timeout:
+      - X-STAINLESS-READ-TIMEOUT-XXX
+      x-stainless-retry-count:
+      - '0'
+      x-stainless-runtime:
+      - CPython
+      x-stainless-runtime-version:
+      - 3.13.0
+    method: POST
+    uri: https://api.openai.com/v1/chat/completions
+  response:
+    body:
+      string: "{\n  \"id\": \"chatcmpl-D2HpStmyOpe9DrthWBlDdMZfVMJ1u\",\n  \"object\":
+        \"chat.completion\",\n  \"created\": 1769437562,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n
+        \ \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\":
+        \"assistant\",\n        \"content\": \"Hi! How can I assist you today?\",\n
+        \       \"refusal\": null,\n        \"annotations\": []\n      },\n      \"logprobs\":
+        null,\n      \"finish_reason\": \"stop\"\n    }\n  ],\n  \"usage\": {\n    \"prompt_tokens\":
+        9,\n    \"completion_tokens\": 9,\n    \"total_tokens\": 18,\n    \"prompt_tokens_details\":
+        {\n      \"cached_tokens\": 0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\":
+        {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\":
+        0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\":
+        \"default\",\n  \"system_fingerprint\": \"fp_29330a9688\"\n}\n"
+    headers:
+      CF-RAY:
+      - CF-RAY-XXX
+      Connection:
+      - keep-alive
+      Content-Type:
+      - application/json
+      Date:
+      - Mon, 26 Jan 2026 14:26:02 GMT
+      Server:
+      - cloudflare
+      Set-Cookie:
+      - SET-COOKIE-XXX
+      Strict-Transport-Security:
+      - STS-XXX
+      Transfer-Encoding:
+      - chunked
+      X-Content-Type-Options:
+      - X-CONTENT-TYPE-XXX
+      access-control-expose-headers:
+      - ACCESS-CONTROL-XXX
+      alt-svc:
+      - h3=":443"; ma=86400
+      cf-cache-status:
+      - DYNAMIC
+      openai-organization:
+      - OPENAI-ORG-XXX
+      openai-processing-ms:
+      - '415'
+      openai-project:
+      - OPENAI-PROJECT-XXX
+      openai-version:
+      - '2020-10-01'
+      x-envoy-upstream-service-time:
+      - '434'
+      x-openai-proxy-wasm:
+      - v0.1
+      x-ratelimit-limit-requests:
+      - X-RATELIMIT-LIMIT-REQUESTS-XXX
+      x-ratelimit-limit-tokens:
+      - X-RATELIMIT-LIMIT-TOKENS-XXX
+      x-ratelimit-remaining-requests:
+      - X-RATELIMIT-REMAINING-REQUESTS-XXX
+      x-ratelimit-remaining-tokens:
+      - X-RATELIMIT-REMAINING-TOKENS-XXX
+      x-ratelimit-reset-requests:
+      - X-RATELIMIT-RESET-REQUESTS-XXX
+      x-ratelimit-reset-tokens:
+      - X-RATELIMIT-RESET-TOKENS-XXX
+      x-request-id:
+      - X-REQUEST-ID-XXX
+    status:
+      code: 200
+      message: OK
+- request:
+    body: '{"messages":[{"role":"user","content":"Say bye"}],"model":"gpt-4o-mini"}'
+    headers:
+      User-Agent:
+      - X-USER-AGENT-XXX
+      accept:
+      - application/json
+      accept-encoding:
+      - ACCEPT-ENCODING-XXX
+      authorization:
+      - AUTHORIZATION-XXX
+      connection:
+      - keep-alive
+      content-length:
+      - '72'
+      content-type:
+      - application/json
+      cookie:
+      - COOKIE-XXX
+      host:
+      - api.openai.com
+      x-stainless-arch:
+      - X-STAINLESS-ARCH-XXX
+      x-stainless-async:
+      - 'false'
+      x-stainless-lang:
+      - python
+      x-stainless-os:
+      - X-STAINLESS-OS-XXX
+      x-stainless-package-version:
+      - 1.83.0
+      x-stainless-read-timeout:
+      - X-STAINLESS-READ-TIMEOUT-XXX
+      x-stainless-retry-count:
+      - '0'
+      x-stainless-runtime:
+      - CPython
+      x-stainless-runtime-version:
+      - 3.13.0
+    method: POST
+    uri: https://api.openai.com/v1/chat/completions
+  response:
+    body:
+      string: "{\n  \"id\": \"chatcmpl-D2HpS1DP0Xd3tmWt5PBincVrdU7yw\",\n  \"object\":
+        \"chat.completion\",\n  \"created\": 1769437562,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n
+        \ \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\":
+        \"assistant\",\n        \"content\": \"Goodbye! If you have more questions
+        in the future, feel free to reach out. Have a great day!\",\n        \"refusal\":
+        null,\n        \"annotations\": []\n      },\n      \"logprobs\": null,\n
+        \     \"finish_reason\": \"stop\"\n    }\n  ],\n  \"usage\": {\n    \"prompt_tokens\":
+        9,\n    \"completion_tokens\": 23,\n    \"total_tokens\": 32,\n    \"prompt_tokens_details\":
+        {\n      \"cached_tokens\": 0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\":
+        {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\":
+        0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\":
+        \"default\",\n  \"system_fingerprint\": \"fp_29330a9688\"\n}\n"
+    headers:
+      CF-RAY:
+      - CF-RAY-XXX
+      Connection:
+      - keep-alive
+      Content-Type:
+      - application/json
+      Date:
+      - Mon, 26 Jan 2026 14:26:03 GMT
+      Server:
+      - cloudflare
+      Strict-Transport-Security:
+      - STS-XXX
+      Transfer-Encoding:
+      - chunked
+      X-Content-Type-Options:
+      - X-CONTENT-TYPE-XXX
+      access-control-expose-headers:
+      - ACCESS-CONTROL-XXX
+      alt-svc:
+      - h3=":443"; ma=86400
+      cf-cache-status:
+      - DYNAMIC
+      openai-organization:
+      - OPENAI-ORG-XXX
+      openai-processing-ms:
+      - '964'
+      openai-project:
+      - OPENAI-PROJECT-XXX
+      openai-version:
+      - '2020-10-01'
+      x-envoy-upstream-service-time:
+      - '979'
+      x-openai-proxy-wasm:
+      - v0.1
+      x-ratelimit-limit-requests:
+      - X-RATELIMIT-LIMIT-REQUESTS-XXX
+      x-ratelimit-limit-tokens:
+      - X-RATELIMIT-LIMIT-TOKENS-XXX
+      x-ratelimit-remaining-requests:
+      - X-RATELIMIT-REMAINING-REQUESTS-XXX
+      x-ratelimit-remaining-tokens:
+      - X-RATELIMIT-REMAINING-TOKENS-XXX
+      x-ratelimit-reset-requests:
+      - X-RATELIMIT-RESET-REQUESTS-XXX
+      x-ratelimit-reset-tokens:
+      - X-RATELIMIT-RESET-TOKENS-XXX
+      x-request-id:
+      - X-REQUEST-ID-XXX
+    status:
+      code: 200
+      message: OK
+version: 1

lib/crewai/tests/cassettes/utilities/test_streaming_chunks_share_call_id_with_call.yaml

@@ -0,0 +1,143 @@
+interactions:
+- request:
+    body: '{"messages":[{"role":"user","content":"Say hi"}],"model":"gpt-4o-mini","stream":true,"stream_options":{"include_usage":true}}'
+    headers:
+      User-Agent:
+      - X-USER-AGENT-XXX
+      accept:
+      - application/json
+      accept-encoding:
+      - ACCEPT-ENCODING-XXX
+      authorization:
+      - AUTHORIZATION-XXX
+      connection:
+      - keep-alive
+      content-length:
+      - '125'
+      content-type:
+      - application/json
+      host:
+      - api.openai.com
+      x-stainless-arch:
+      - X-STAINLESS-ARCH-XXX
+      x-stainless-async:
+      - 'false'
+      x-stainless-lang:
+      - python
+      x-stainless-os:
+      - X-STAINLESS-OS-XXX
+      x-stainless-package-version:
+      - 1.83.0
+      x-stainless-read-timeout:
+      - X-STAINLESS-READ-TIMEOUT-XXX
+      x-stainless-retry-count:
+      - '0'
+      x-stainless-runtime:
+      - CPython
+      x-stainless-runtime-version:
+      - 3.13.0
+    method: POST
+    uri: https://api.openai.com/v1/chat/completions
+  response:
+    body:
+      string: 'data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"role":"assistant","content":"","refusal":null},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"rVIyGQF2E"}
+
+
+        data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"content":"Hi"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"ZGVqV7ZDm"}
+
+
+        data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"content":"!"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"vnfm7IxlIB"}
+
+
+        data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"content":"
+        How"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"o8F35ZZ"}
+
+
+        data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"content":"
+        can"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"kiBzGe3"}
+
+
+        data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"content":"
+        I"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"cbGT2RWgx"}
+
+
+        data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"content":"
+        assist"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"DtxR"}
+
+
+        data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"content":"
+        you"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"6y6Co8J"}
+
+
+        data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"content":"
+        today"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"SZOmm"}
+
+
+        data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{"content":"?"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"s9Bc0HqlPg"}
+
+
+        data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[{"index":0,"delta":{},"logprobs":null,"finish_reason":"stop"}],"usage":null,"obfuscation":"u9aar"}
+
+
+        data: {"id":"chatcmpl-D2HpUGTvIFKBsR9Xd6XRT4AuFXzbz","object":"chat.completion.chunk","created":1769437564,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_29330a9688","choices":[],"usage":{"prompt_tokens":9,"completion_tokens":9,"total_tokens":18,"prompt_tokens_details":{"cached_tokens":0,"audio_tokens":0},"completion_tokens_details":{"reasoning_tokens":0,"audio_tokens":0,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"obfuscation":"5hudm8ySqh39"}
+
+
+        data: [DONE]
+
+
+        '
+    headers:
+      CF-RAY:
+      - CF-RAY-XXX
+      Connection:
+      - keep-alive
+      Content-Type:
+      - text/event-stream; charset=utf-8
+      Date:
+      - Mon, 26 Jan 2026 14:26:04 GMT
+      Server:
+      - cloudflare
+      Set-Cookie:
+      - SET-COOKIE-XXX
+      Strict-Transport-Security:
+      - STS-XXX
+      Transfer-Encoding:
+      - chunked
+      X-Content-Type-Options:
+      - X-CONTENT-TYPE-XXX
+      access-control-expose-headers:
+      - ACCESS-CONTROL-XXX
+      alt-svc:
+      - h3=":443"; ma=86400
+      cf-cache-status:
+      - DYNAMIC
+      openai-organization:
+      - OPENAI-ORG-XXX
+      openai-processing-ms:
+      - '260'
+      openai-project:
+      - OPENAI-PROJECT-XXX
+      openai-version:
+      - '2020-10-01'
+      x-envoy-upstream-service-time:
+      - '275'
+      x-openai-proxy-wasm:
+      - v0.1
+      x-ratelimit-limit-requests:
+      - X-RATELIMIT-LIMIT-REQUESTS-XXX
+      x-ratelimit-limit-tokens:
+      - X-RATELIMIT-LIMIT-TOKENS-XXX
+      x-ratelimit-remaining-requests:
+      - X-RATELIMIT-REMAINING-REQUESTS-XXX
+      x-ratelimit-remaining-tokens:
+      - X-RATELIMIT-REMAINING-TOKENS-XXX
+      x-ratelimit-reset-requests:
+      - X-RATELIMIT-RESET-REQUESTS-XXX
+      x-ratelimit-reset-tokens:
+      - X-RATELIMIT-RESET-TOKENS-XXX
+      x-request-id:
+      - X-REQUEST-ID-XXX
+    status:
+      code: 200
+      message: OK
+version: 1

lib/crewai/tests/cli/test_create_crew.py

@@ -296,6 +296,23 @@ def test_create_folder_structure_folder_name_validation():
                 shutil.rmtree(folder_path)
 
 
+def test_create_folder_structure_rejects_reserved_names():
+    """Test that reserved script names are rejected to prevent pyproject.toml conflicts."""
+    with tempfile.TemporaryDirectory() as temp_dir:
+        reserved_names = ["test", "train", "replay", "run_crew", "run_with_trigger"]
+
+        for reserved_name in reserved_names:
+            with pytest.raises(ValueError, match="which is reserved"):
+                create_folder_structure(reserved_name, parent_folder=temp_dir)
+
+            with pytest.raises(ValueError, match="which is reserved"):
+                create_folder_structure(f"{reserved_name}/", parent_folder=temp_dir)
+
+            capitalized = reserved_name.capitalize()
+            with pytest.raises(ValueError, match="which is reserved"):
+                create_folder_structure(capitalized, parent_folder=temp_dir)
+
+
 @mock.patch("crewai.cli.create_crew.create_folder_structure")
 @mock.patch("crewai.cli.create_crew.copy_template")
 @mock.patch("crewai.cli.create_crew.load_env_vars")

lib/crewai/tests/cli/test_plus_api.py

@@ -1,6 +1,8 @@
 import os
 import unittest
-from unittest.mock import ANY, MagicMock, patch
+from unittest.mock import ANY, AsyncMock, MagicMock, patch
+
+import pytest
 
 from crewai.cli.plus_api import PlusAPI
 
@@ -68,37 +70,6 @@ class TestPlusAPI(unittest.TestCase):
         )
         self.assertEqual(response, mock_response)
 
-    @patch("crewai.cli.plus_api.PlusAPI._make_request")
-    def test_get_agent(self, mock_make_request):
-        mock_response = MagicMock()
-        mock_make_request.return_value = mock_response
-
-        response = self.api.get_agent("test_agent_handle")
-        mock_make_request.assert_called_once_with(
-            "GET", "/crewai_plus/api/v1/agents/test_agent_handle"
-        )
-        self.assertEqual(response, mock_response)
-
-    @patch("crewai.cli.plus_api.Settings")
-    @patch("requests.Session.request")
-    def test_get_agent_with_org_uuid(self, mock_make_request, mock_settings_class):
-        mock_settings = MagicMock()
-        mock_settings.org_uuid = self.org_uuid
-        mock_settings.enterprise_base_url = os.getenv('CREWAI_PLUS_URL')
-        mock_settings_class.return_value = mock_settings
-        # re-initialize Client
-        self.api = PlusAPI(self.api_key)
-
-        mock_response = MagicMock()
-        mock_make_request.return_value = mock_response
-
-        response = self.api.get_agent("test_agent_handle")
-
-        self.assert_request_with_org_id(
-            mock_make_request, "GET", "/crewai_plus/api/v1/agents/test_agent_handle"
-        )
-        self.assertEqual(response, mock_response)
-
     @patch("crewai.cli.plus_api.PlusAPI._make_request")
     def test_get_tool(self, mock_make_request):
         mock_response = MagicMock()
@@ -338,3 +309,49 @@ class TestPlusAPI(unittest.TestCase):
             custom_api.base_url,
             "https://custom-url-from-env.com",
         )
+
+
+@pytest.mark.asyncio
+@patch("httpx.AsyncClient")
+async def test_get_agent(mock_async_client_class):
+    api = PlusAPI("test_api_key")
+    mock_response = MagicMock()
+    mock_client_instance = AsyncMock()
+    mock_client_instance.get.return_value = mock_response
+    mock_async_client_class.return_value.__aenter__.return_value = mock_client_instance
+
+    response = await api.get_agent("test_agent_handle")
+
+    mock_client_instance.get.assert_called_once_with(
+        f"{api.base_url}/crewai_plus/api/v1/agents/test_agent_handle",
+        headers=api.headers,
+    )
+    assert response == mock_response
+
+
+@pytest.mark.asyncio
+@patch("httpx.AsyncClient")
+@patch("crewai.cli.plus_api.Settings")
+async def test_get_agent_with_org_uuid(mock_settings_class, mock_async_client_class):
+    org_uuid = "test-org-uuid"
+    mock_settings = MagicMock()
+    mock_settings.org_uuid = org_uuid
+    mock_settings.enterprise_base_url = os.getenv("CREWAI_PLUS_URL")
+    mock_settings_class.return_value = mock_settings
+
+    api = PlusAPI("test_api_key")
+
+    mock_response = MagicMock()
+    mock_client_instance = AsyncMock()
+    mock_client_instance.get.return_value = mock_response
+    mock_async_client_class.return_value.__aenter__.return_value = mock_client_instance
+
+    response = await api.get_agent("test_agent_handle")
+
+    mock_client_instance.get.assert_called_once_with(
+        f"{api.base_url}/crewai_plus/api/v1/agents/test_agent_handle",
+        headers=api.headers,
+    )
+    assert "X-Crewai-Organization-Id" in api.headers
+    assert api.headers["X-Crewai-Organization-Id"] == org_uuid
+    assert response == mock_response

lib/crewai/tests/cli/test_version.py

@@ -1,10 +1,20 @@
 """Test for version management."""
 
+from datetime import datetime, timedelta
+from pathlib import Path
+from unittest.mock import MagicMock, patch
+
 from crewai import __version__
-from crewai.cli.version import get_crewai_version
+from crewai.cli.version import (
+    _get_cache_file,
+    _is_cache_valid,
+    get_crewai_version,
+    get_latest_version_from_pypi,
+    is_newer_version_available,
+)
 
 
-def test_dynamic_versioning_consistency():
+def test_dynamic_versioning_consistency() -> None:
     """Test that dynamic versioning provides consistent version across all access methods."""
     cli_version = get_crewai_version()
     package_version = __version__
@@ -15,3 +25,186 @@ def test_dynamic_versioning_consistency():
     # Version should not be empty
     assert package_version is not None
     assert len(package_version.strip()) > 0
+
+
+class TestVersionChecking:
+    """Test version checking utilities."""
+
+    def test_get_crewai_version(self) -> None:
+        """Test getting current crewai version."""
+        version = get_crewai_version()
+        assert isinstance(version, str)
+        assert len(version) > 0
+
+    def test_get_cache_file(self) -> None:
+        """Test cache file path generation."""
+        cache_file = _get_cache_file()
+        assert isinstance(cache_file, Path)
+        assert cache_file.name == "version_cache.json"
+
+    def test_is_cache_valid_with_fresh_cache(self) -> None:
+        """Test cache validation with fresh cache."""
+        cache_data = {"timestamp": datetime.now().isoformat(), "version": "1.0.0"}
+        assert _is_cache_valid(cache_data) is True
+
+    def test_is_cache_valid_with_stale_cache(self) -> None:
+        """Test cache validation with stale cache."""
+        old_time = datetime.now() - timedelta(hours=25)
+        cache_data = {"timestamp": old_time.isoformat(), "version": "1.0.0"}
+        assert _is_cache_valid(cache_data) is False
+
+    def test_is_cache_valid_with_missing_timestamp(self) -> None:
+        """Test cache validation with missing timestamp."""
+        cache_data = {"version": "1.0.0"}
+        assert _is_cache_valid(cache_data) is False
+
+    @patch("crewai.cli.version.Path.exists")
+    @patch("crewai.cli.version.request.urlopen")
+    def test_get_latest_version_from_pypi_success(
+        self, mock_urlopen: MagicMock, mock_exists: MagicMock
+    ) -> None:
+        """Test successful PyPI version fetch."""
+        # Mock cache not existing to force fetch from PyPI
+        mock_exists.return_value = False
+
+        mock_response = MagicMock()
+        mock_response.read.return_value = b'{"info": {"version": "2.0.0"}}'
+        mock_urlopen.return_value.__enter__.return_value = mock_response
+
+        version = get_latest_version_from_pypi()
+        assert version == "2.0.0"
+
+    @patch("crewai.cli.version.Path.exists")
+    @patch("crewai.cli.version.request.urlopen")
+    def test_get_latest_version_from_pypi_failure(
+        self, mock_urlopen: MagicMock, mock_exists: MagicMock
+    ) -> None:
+        """Test PyPI version fetch failure."""
+        from urllib.error import URLError
+
+        # Mock cache not existing to force fetch from PyPI
+        mock_exists.return_value = False
+
+        mock_urlopen.side_effect = URLError("Network error")
+
+        version = get_latest_version_from_pypi()
+        assert version is None
+
+    @patch("crewai.cli.version.get_crewai_version")
+    @patch("crewai.cli.version.get_latest_version_from_pypi")
+    def test_is_newer_version_available_true(
+        self, mock_latest: MagicMock, mock_current: MagicMock
+    ) -> None:
+        """Test when newer version is available."""
+        mock_current.return_value = "1.0.0"
+        mock_latest.return_value = "2.0.0"
+
+        is_newer, current, latest = is_newer_version_available()
+        assert is_newer is True
+        assert current == "1.0.0"
+        assert latest == "2.0.0"
+
+    @patch("crewai.cli.version.get_crewai_version")
+    @patch("crewai.cli.version.get_latest_version_from_pypi")
+    def test_is_newer_version_available_false(
+        self, mock_latest: MagicMock, mock_current: MagicMock
+    ) -> None:
+        """Test when no newer version is available."""
+        mock_current.return_value = "2.0.0"
+        mock_latest.return_value = "2.0.0"
+
+        is_newer, current, latest = is_newer_version_available()
+        assert is_newer is False
+        assert current == "2.0.0"
+        assert latest == "2.0.0"
+
+    @patch("crewai.cli.version.get_crewai_version")
+    @patch("crewai.cli.version.get_latest_version_from_pypi")
+    def test_is_newer_version_available_with_none_latest(
+        self, mock_latest: MagicMock, mock_current: MagicMock
+    ) -> None:
+        """Test when PyPI fetch fails."""
+        mock_current.return_value = "1.0.0"
+        mock_latest.return_value = None
+
+        is_newer, current, latest = is_newer_version_available()
+        assert is_newer is False
+        assert current == "1.0.0"
+        assert latest is None
+
+
+class TestConsoleFormatterVersionCheck:
+    """Test version check display in ConsoleFormatter."""
+
+    @patch("crewai.events.utils.console_formatter.is_newer_version_available")
+    @patch.dict("os.environ", {"CI": ""})
+    def test_version_message_shows_when_update_available_and_verbose(
+        self, mock_check: MagicMock
+    ) -> None:
+        """Test version message shows when update available and verbose enabled."""
+        from crewai.events.utils.console_formatter import ConsoleFormatter
+
+        mock_check.return_value = (True, "1.0.0", "2.0.0")
+
+        formatter = ConsoleFormatter(verbose=True)
+        with patch.object(formatter.console, "print") as mock_print:
+            formatter._show_version_update_message_if_needed()
+            assert mock_print.call_count == 2
+
+    @patch("crewai.events.utils.console_formatter.is_newer_version_available")
+    def test_version_message_hides_when_verbose_false(
+        self, mock_check: MagicMock
+    ) -> None:
+        """Test version message hidden when verbose disabled."""
+        from crewai.events.utils.console_formatter import ConsoleFormatter
+
+        mock_check.return_value = (True, "1.0.0", "2.0.0")
+
+        formatter = ConsoleFormatter(verbose=False)
+        with patch.object(formatter.console, "print") as mock_print:
+            formatter._show_version_update_message_if_needed()
+            mock_print.assert_not_called()
+
+    @patch("crewai.events.utils.console_formatter.is_newer_version_available")
+    def test_version_message_hides_when_no_update_available(
+        self, mock_check: MagicMock
+    ) -> None:
+        """Test version message hidden when no update available."""
+        from crewai.events.utils.console_formatter import ConsoleFormatter
+
+        mock_check.return_value = (False, "2.0.0", "2.0.0")
+
+        formatter = ConsoleFormatter(verbose=True)
+        with patch.object(formatter.console, "print") as mock_print:
+            formatter._show_version_update_message_if_needed()
+            mock_print.assert_not_called()
+
+    @patch("crewai.events.utils.console_formatter.is_newer_version_available")
+    @patch.dict("os.environ", {"CI": "true"})
+    def test_version_message_hides_in_ci_environment(
+        self, mock_check: MagicMock
+    ) -> None:
+        """Test version message hidden when running in CI/CD."""
+        from crewai.events.utils.console_formatter import ConsoleFormatter
+
+        mock_check.return_value = (True, "1.0.0", "2.0.0")
+
+        formatter = ConsoleFormatter(verbose=True)
+        with patch.object(formatter.console, "print") as mock_print:
+            formatter._show_version_update_message_if_needed()
+            mock_print.assert_not_called()
+
+    @patch("crewai.events.utils.console_formatter.is_newer_version_available")
+    @patch.dict("os.environ", {"CI": "1"})
+    def test_version_message_hides_in_ci_environment_with_numeric_value(
+        self, mock_check: MagicMock
+    ) -> None:
+        """Test version message hidden when CI=1."""
+        from crewai.events.utils.console_formatter import ConsoleFormatter
+
+        mock_check.return_value = (True, "1.0.0", "2.0.0")
+
+        formatter = ConsoleFormatter(verbose=True)
+        with patch.object(formatter.console, "print") as mock_print:
+            formatter._show_version_update_message_if_needed()
+            mock_print.assert_not_called()

lib/crewai/tests/events/test_event_context.py

@@ -177,4 +177,40 @@ class TestTriggeredByScope:
                 raise ValueError("test error")
         except ValueError:
             pass
-        assert get_triggering_event_id() is None
+        assert get_triggering_event_id() is None
+
+
+def test_agent_scope_preserved_after_tool_error_event() -> None:
+    from crewai.events import crewai_event_bus
+    from crewai.events.types.tool_usage_events import (
+        ToolUsageErrorEvent,
+        ToolUsageStartedEvent,
+    )
+
+    push_event_scope("crew-1", "crew_kickoff_started")
+    push_event_scope("task-1", "task_started")
+    push_event_scope("agent-1", "agent_execution_started")
+
+    crewai_event_bus.emit(
+        None,
+        ToolUsageStartedEvent(
+            tool_name="test_tool",
+            tool_args={},
+            agent_key="test_agent",
+        )
+    )
+
+    crewai_event_bus.emit(
+        None,
+        ToolUsageErrorEvent(
+            tool_name="test_tool",
+            tool_args={},
+            agent_key="test_agent",
+            error=ValueError("test error"),
+        )
+    )
+
+    crewai_event_bus.flush()
+
+    assert get_current_parent_id() == "agent-1"
+

lib/crewai/tests/llms/anthropic/test_anthropic.py

@@ -45,85 +45,6 @@ def test_anthropic_completion_is_used_when_claude_provider():
 
 
 
-def test_anthropic_tool_use_conversation_flow():
-    """
-    Test that the Anthropic completion properly handles tool use conversation flow
-    """
-    from unittest.mock import Mock, patch
-    from crewai.llms.providers.anthropic.completion import AnthropicCompletion
-    from anthropic.types.tool_use_block import ToolUseBlock
-
-    # Create AnthropicCompletion instance
-    completion = AnthropicCompletion(model="claude-3-5-sonnet-20241022")
-
-    # Mock tool function
-    def mock_weather_tool(location: str) -> str:
-        return f"The weather in {location} is sunny and 75°F"
-
-    available_functions = {"get_weather": mock_weather_tool}
-
-    # Mock the Anthropic client responses
-    with patch.object(completion.client.messages, 'create') as mock_create:
-        # Mock initial response with tool use - need to properly mock ToolUseBlock
-        mock_tool_use = Mock(spec=ToolUseBlock)
-        mock_tool_use.type = "tool_use"
-        mock_tool_use.id = "tool_123"
-        mock_tool_use.name = "get_weather"
-        mock_tool_use.input = {"location": "San Francisco"}
-
-        mock_initial_response = Mock()
-        mock_initial_response.content = [mock_tool_use]
-        mock_initial_response.usage = Mock()
-        mock_initial_response.usage.input_tokens = 100
-        mock_initial_response.usage.output_tokens = 50
-
-        # Mock final response after tool result - properly mock text content
-        mock_text_block = Mock()
-        mock_text_block.type = "text"
-        # Set the text attribute as a string, not another Mock
-        mock_text_block.configure_mock(text="Based on the weather data, it's a beautiful day in San Francisco with sunny skies and 75°F temperature.")
-
-        mock_final_response = Mock()
-        mock_final_response.content = [mock_text_block]
-        mock_final_response.usage = Mock()
-        mock_final_response.usage.input_tokens = 150
-        mock_final_response.usage.output_tokens = 75
-
-        # Configure mock to return different responses on successive calls
-        mock_create.side_effect = [mock_initial_response, mock_final_response]
-
-        # Test the call
-        messages = [{"role": "user", "content": "What's the weather like in San Francisco?"}]
-        result = completion.call(
-            messages=messages,
-            available_functions=available_functions
-        )
-
-        # Verify the result contains the final response
-        assert "beautiful day in San Francisco" in result
-        assert "sunny skies" in result
-        assert "75°F" in result
-
-        # Verify that two API calls were made (initial + follow-up)
-        assert mock_create.call_count == 2
-
-        # Verify the second call includes tool results
-        second_call_args = mock_create.call_args_list[1][1]  # kwargs of second call
-        messages_in_second_call = second_call_args["messages"]
-
-        # Should have original user message + assistant tool use + user tool result
-        assert len(messages_in_second_call) == 3
-        assert messages_in_second_call[0]["role"] == "user"
-        assert messages_in_second_call[1]["role"] == "assistant"
-        assert messages_in_second_call[2]["role"] == "user"
-
-        # Verify tool result format
-        tool_result = messages_in_second_call[2]["content"][0]
-        assert tool_result["type"] == "tool_result"
-        assert tool_result["tool_use_id"] == "tool_123"
-        assert "sunny and 75°F" in tool_result["content"]
-
-
 def test_anthropic_completion_module_is_imported():
     """
     Test that the completion module is properly imported when using Anthropic provider
@@ -874,6 +795,125 @@ def test_anthropic_function_calling():
 # =============================================================================
 
 
+@pytest.mark.vcr(filter_headers=["authorization", "x-api-key"])
+def test_anthropic_tool_execution_with_available_functions():
+    """
+    Test that Anthropic provider correctly executes tools when available_functions is provided.
+
+    This specifically tests the fix for double llm_call_completed emission - when
+    available_functions is provided, _handle_tool_execution is called which already
+    emits llm_call_completed, so the caller should not emit it again.
+
+    The test verifies:
+    1. The tool is called with correct arguments
+    2. The tool result is returned directly (not wrapped in conversation)
+    3. The result is valid JSON matching the tool output format
+    """
+    import json
+
+    llm = LLM(model="anthropic/claude-3-5-haiku-20241022")
+
+    # Simple tool that returns a formatted string
+    def create_reasoning_plan(plan: str, steps: list, ready: bool) -> str:
+        """Create a reasoning plan with steps."""
+        return json.dumps({"plan": plan, "steps": steps, "ready": ready})
+
+    tools = [
+        {
+            "name": "create_reasoning_plan",
+            "description": "Create a structured reasoning plan for completing a task",
+            "input_schema": {
+                "type": "object",
+                "properties": {
+                    "plan": {
+                        "type": "string",
+                        "description": "High-level plan description"
+                    },
+                    "steps": {
+                        "type": "array",
+                        "items": {"type": "object"},
+                        "description": "List of steps to execute"
+                    },
+                    "ready": {
+                        "type": "boolean",
+                        "description": "Whether the plan is ready to execute"
+                    }
+                },
+                "required": ["plan", "steps", "ready"]
+            }
+        }
+    ]
+
+    result = llm.call(
+        messages=[{"role": "user", "content": "Create a simple plan to say hello. Use the create_reasoning_plan tool."}],
+        tools=tools,
+        available_functions={"create_reasoning_plan": create_reasoning_plan}
+    )
+
+    # Verify result is valid JSON from the tool
+    assert result is not None
+    assert isinstance(result, str)
+
+    # Parse the result to verify it's valid JSON
+    parsed_result = json.loads(result)
+    assert "plan" in parsed_result
+    assert "steps" in parsed_result
+    assert "ready" in parsed_result
+
+
+@pytest.mark.vcr(filter_headers=["authorization", "x-api-key"])
+def test_anthropic_tool_execution_returns_tool_result_directly():
+    """
+    Test that when available_functions is provided, the tool result is returned directly
+    without additional LLM conversation (matching OpenAI behavior for reasoning_handler).
+    """
+    llm = LLM(model="anthropic/claude-3-5-haiku-20241022")
+
+    call_count = 0
+
+    def simple_calculator(operation: str, a: int, b: int) -> str:
+        """Perform a simple calculation."""
+        nonlocal call_count
+        call_count += 1
+        if operation == "add":
+            return str(a + b)
+        elif operation == "multiply":
+            return str(a * b)
+        return "Unknown operation"
+
+    tools = [
+        {
+            "name": "simple_calculator",
+            "description": "Perform simple math operations",
+            "input_schema": {
+                "type": "object",
+                "properties": {
+                    "operation": {
+                        "type": "string",
+                        "enum": ["add", "multiply"],
+                        "description": "The operation to perform"
+                    },
+                    "a": {"type": "integer", "description": "First number"},
+                    "b": {"type": "integer", "description": "Second number"}
+                },
+                "required": ["operation", "a", "b"]
+            }
+        }
+    ]
+
+    result = llm.call(
+        messages=[{"role": "user", "content": "Calculate 5 + 3 using the simple_calculator tool with operation 'add'."}],
+        tools=tools,
+        available_functions={"simple_calculator": simple_calculator}
+    )
+
+    # Tool should have been called exactly once
+    assert call_count == 1, f"Expected tool to be called once, got {call_count}"
+
+    # Result should be the direct tool output
+    assert result == "8", f"Expected '8' but got '{result}'"
+
+
 @pytest.mark.vcr()
 def test_anthropic_agent_kickoff_structured_output_without_tools():
     """

lib/crewai/tests/memory/test_external_memory.py

@@ -308,6 +308,7 @@ def test_external_memory_search_events(
         "parent_event_id": None,
         "previous_event_id": ANY,
         "triggered_by_event_id": None,
+        "started_event_id": ANY,
         "emission_sequence": ANY,
         "query": "test value",
         "limit": 3,
@@ -330,6 +331,7 @@ def test_external_memory_search_events(
         "parent_event_id": ANY,
         "previous_event_id": ANY,
         "triggered_by_event_id": None,
+        "started_event_id": ANY,
         "emission_sequence": ANY,
         "query": "test value",
         "results": [],
@@ -390,6 +392,7 @@ def test_external_memory_save_events(
         "parent_event_id": None,
         "previous_event_id": ANY,
         "triggered_by_event_id": None,
+        "started_event_id": ANY,
         "emission_sequence": ANY,
         "value": "saving value",
         "metadata": {"task": "test_task"},
@@ -411,6 +414,7 @@ def test_external_memory_save_events(
         "parent_event_id": ANY,
         "previous_event_id": ANY,
         "triggered_by_event_id": None,
+        "started_event_id": ANY,
         "emission_sequence": ANY,
         "value": "saving value",
         "metadata": {"task": "test_task"},

lib/crewai/tests/memory/test_long_term_memory.py

@@ -74,6 +74,7 @@ def test_long_term_memory_save_events(long_term_memory):
         "parent_event_id": None,
         "previous_event_id": ANY,
         "triggered_by_event_id": None,
+        "started_event_id": ANY,
         "emission_sequence": ANY,
         "value": "test_task",
         "metadata": {"task": "test_task", "quality": 0.5},
@@ -94,6 +95,7 @@ def test_long_term_memory_save_events(long_term_memory):
         "parent_event_id": None,
         "previous_event_id": ANY,
         "triggered_by_event_id": None,
+        "started_event_id": ANY,
         "emission_sequence": ANY,
         "value": "test_task",
         "metadata": {
@@ -153,6 +155,7 @@ def test_long_term_memory_search_events(long_term_memory):
         "parent_event_id": None,
         "previous_event_id": ANY,
         "triggered_by_event_id": None,
+        "started_event_id": ANY,
         "emission_sequence": ANY,
         "query": "test query",
         "limit": 5,
@@ -175,6 +178,7 @@ def test_long_term_memory_search_events(long_term_memory):
         "parent_event_id": ANY,
         "previous_event_id": ANY,
         "triggered_by_event_id": None,
+        "started_event_id": ANY,
         "emission_sequence": ANY,
         "query": "test query",
         "results": None,

lib/crewai/tests/memory/test_short_term_memory.py

@@ -85,6 +85,7 @@ def test_short_term_memory_search_events(short_term_memory):
         "parent_event_id": None,
         "previous_event_id": ANY,
         "triggered_by_event_id": None,
+        "started_event_id": ANY,
         "emission_sequence": ANY,
         "query": "test value",
         "limit": 3,
@@ -107,6 +108,7 @@ def test_short_term_memory_search_events(short_term_memory):
         "parent_event_id": None,
         "previous_event_id": ANY,
         "triggered_by_event_id": None,
+        "started_event_id": ANY,
         "emission_sequence": ANY,
         "query": "test value",
         "results": [],
@@ -164,6 +166,7 @@ def test_short_term_memory_save_events(short_term_memory):
         "parent_event_id": None,
         "previous_event_id": ANY,
         "triggered_by_event_id": None,
+        "started_event_id": ANY,
         "emission_sequence": ANY,
         "value": "test value",
         "metadata": {"task": "test_task"},
@@ -185,6 +188,7 @@ def test_short_term_memory_save_events(short_term_memory):
         "parent_event_id": None,
         "previous_event_id": ANY,
         "triggered_by_event_id": None,
+        "started_event_id": ANY,
         "emission_sequence": ANY,
         "value": "test value",
         "metadata": {"task": "test_task"},

lib/crewai/tests/test_crew.py

@@ -4520,7 +4520,7 @@ def test_crew_copy_with_memory():
         pytest.fail(f"Copying crew raised an unexpected exception: {e}")
 
 
-def test_sets_parent_flow_when_using_crewbase_pattern_inside_flow():
+def test_sets_flow_context_when_using_crewbase_pattern_inside_flow():
     @CrewBase
     class TestCrew:
         agents_config = None
@@ -4582,10 +4582,11 @@ def test_sets_parent_flow_when_using_crewbase_pattern_inside_flow():
     flow.kickoff()
 
     assert captured_crew is not None
-    assert captured_crew.parent_flow is flow
+    assert captured_crew._flow_id == flow.flow_id  # type: ignore[attr-defined]
+    assert captured_crew._request_id == flow.flow_id  # type: ignore[attr-defined]
 
 
-def test_sets_parent_flow_when_outside_flow(researcher, writer):
+def test_sets_flow_context_when_outside_flow(researcher, writer):
     crew = Crew(
         agents=[researcher, writer],
         process=Process.sequential,
@@ -4594,11 +4595,12 @@ def test_sets_parent_flow_when_outside_flow(researcher, writer):
             Task(description="Task 2", expected_output="output", agent=writer),
         ],
     )
-    assert crew.parent_flow is None
+    assert not hasattr(crew, "_flow_id")
+    assert not hasattr(crew, "_request_id")
 
 
 @pytest.mark.vcr()
-def test_sets_parent_flow_when_inside_flow(researcher, writer):
+def test_sets_flow_context_when_inside_flow(researcher, writer):
     class MyFlow(Flow):
         @start()
         def start(self):
@@ -4615,7 +4617,8 @@ def test_sets_parent_flow_when_inside_flow(researcher, writer):
 
     flow = MyFlow()
     result = flow.kickoff()
-    assert result.parent_flow is flow
+    assert result._flow_id == flow.flow_id  # type: ignore[attr-defined]
+    assert result._request_id == flow.flow_id  # type: ignore[attr-defined]
 
 
 def test_reset_knowledge_with_no_crew_knowledge(researcher, writer):

lib/crewai/tests/test_flow_human_input_integration.py

@@ -2,6 +2,7 @@ from unittest.mock import MagicMock, patch
 
 import pytest
 from crewai.events.event_listener import event_listener
+from crewai.core.providers.human_input import SyncHumanInputProvider
 
 
 class TestFlowHumanInputIntegration:
@@ -24,14 +25,9 @@ class TestFlowHumanInputIntegration:
     @patch("builtins.input", return_value="")
     def test_human_input_pauses_flow_updates(self, mock_input):
         """Test that human input pauses Flow status updates."""
-        from crewai.agents.agent_builder.base_agent_executor_mixin import (
-            CrewAgentExecutorMixin,
-        )
-
-        executor = CrewAgentExecutorMixin()
-        executor.crew = MagicMock()
-        executor.crew._train = False
-        executor._printer = MagicMock()
+        provider = SyncHumanInputProvider()
+        crew = MagicMock()
+        crew._train = False
 
         formatter = event_listener.formatter
 
@@ -39,7 +35,7 @@ class TestFlowHumanInputIntegration:
             patch.object(formatter, "pause_live_updates") as mock_pause,
             patch.object(formatter, "resume_live_updates") as mock_resume,
         ):
-            result = executor._ask_human_input("Test result")
+            result = provider._prompt_input(crew)
 
             mock_pause.assert_called_once()
             mock_resume.assert_called_once()
@@ -49,14 +45,9 @@ class TestFlowHumanInputIntegration:
     @patch("builtins.input", side_effect=["feedback", ""])
     def test_multiple_human_input_rounds(self, mock_input):
         """Test multiple rounds of human input with Flow status management."""
-        from crewai.agents.agent_builder.base_agent_executor_mixin import (
-            CrewAgentExecutorMixin,
-        )
-
-        executor = CrewAgentExecutorMixin()
-        executor.crew = MagicMock()
-        executor.crew._train = False
-        executor._printer = MagicMock()
+        provider = SyncHumanInputProvider()
+        crew = MagicMock()
+        crew._train = False
 
         formatter = event_listener.formatter
 
@@ -75,10 +66,10 @@ class TestFlowHumanInputIntegration:
                 formatter, "resume_live_updates", side_effect=track_resume
             ),
         ):
-            result1 = executor._ask_human_input("Test result 1")
+            result1 = provider._prompt_input(crew)
             assert result1 == "feedback"
 
-            result2 = executor._ask_human_input("Test result 2")
+            result2 = provider._prompt_input(crew)
             assert result2 == ""
 
             assert len(pause_calls) == 2
@@ -103,14 +94,9 @@ class TestFlowHumanInputIntegration:
 
     def test_pause_resume_exception_handling(self):
         """Test that resume is called even if exception occurs during human input."""
-        from crewai.agents.agent_builder.base_agent_executor_mixin import (
-            CrewAgentExecutorMixin,
-        )
-
-        executor = CrewAgentExecutorMixin()
-        executor.crew = MagicMock()
-        executor.crew._train = False
-        executor._printer = MagicMock()
+        provider = SyncHumanInputProvider()
+        crew = MagicMock()
+        crew._train = False
 
         formatter = event_listener.formatter
 
@@ -122,21 +108,16 @@ class TestFlowHumanInputIntegration:
             ),
         ):
             with pytest.raises(KeyboardInterrupt):
-                executor._ask_human_input("Test result")
+                provider._prompt_input(crew)
 
             mock_pause.assert_called_once()
             mock_resume.assert_called_once()
 
     def test_training_mode_human_input(self):
         """Test human input in training mode."""
-        from crewai.agents.agent_builder.base_agent_executor_mixin import (
-            CrewAgentExecutorMixin,
-        )
-
-        executor = CrewAgentExecutorMixin()
-        executor.crew = MagicMock()
-        executor.crew._train = True
-        executor._printer = MagicMock()
+        provider = SyncHumanInputProvider()
+        crew = MagicMock()
+        crew._train = True
 
         formatter = event_listener.formatter
 
@@ -146,7 +127,7 @@ class TestFlowHumanInputIntegration:
             patch.object(formatter.console, "print") as mock_console_print,
             patch("builtins.input", return_value="training feedback"),
         ):
-            result = executor._ask_human_input("Test result")
+            result = provider._prompt_input(crew)
 
             mock_pause.assert_called_once()
             mock_resume.assert_called_once()
@@ -161,4 +142,4 @@ class TestFlowHumanInputIntegration:
                 for call in call_args
                 if call[0]
             )
-            assert training_panel_found
+            assert training_panel_found

lib/crewai/tests/test_human_feedback_integration.py

@@ -157,6 +157,176 @@ class TestMultiStepFlows:
 
         assert execution_order == ["generate", "review", "finalize"]
 
+    def test_chained_router_feedback_steps(self):
+        """Test that a router outcome can trigger another router method.
+
+        Regression test: @listen("outcome") combined with @human_feedback(emit=...)
+        creates a method that is both a listener and a router. The flow must find
+        and execute it when the upstream router emits the matching outcome.
+        """
+        execution_order: list[str] = []
+
+        class ChainedRouterFlow(Flow):
+            @start()
+            @human_feedback(
+                message="First review:",
+                emit=["approved", "rejected"],
+                llm="gpt-4o-mini",
+            )
+            def draft(self):
+                execution_order.append("draft")
+                return "draft content"
+
+            @listen("approved")
+            @human_feedback(
+                message="Final review:",
+                emit=["publish", "revise"],
+                llm="gpt-4o-mini",
+            )
+            def final_review(self, prev: HumanFeedbackResult):
+                execution_order.append("final_review")
+                return "final content"
+
+            @listen("rejected")
+            def on_rejected(self, prev: HumanFeedbackResult):
+                execution_order.append("on_rejected")
+                return "rejected"
+
+            @listen("publish")
+            def on_publish(self, prev: HumanFeedbackResult):
+                execution_order.append("on_publish")
+                return "published"
+
+            @listen("revise")
+            def on_revise(self, prev: HumanFeedbackResult):
+                execution_order.append("on_revise")
+                return "revised"
+
+        flow = ChainedRouterFlow()
+
+        with (
+            patch.object(
+                flow,
+                "_request_human_feedback",
+                side_effect=["looks good", "ship it"],
+            ),
+            patch.object(
+                flow,
+                "_collapse_to_outcome",
+                side_effect=["approved", "publish"],
+            ),
+        ):
+            result = flow.kickoff()
+
+        assert execution_order == ["draft", "final_review", "on_publish"]
+        assert result == "published"
+        assert len(flow.human_feedback_history) == 2
+        assert flow.human_feedback_history[0].outcome == "approved"
+        assert flow.human_feedback_history[1].outcome == "publish"
+
+    def test_chained_router_rejected_path(self):
+        """Test that a start-router outcome routes to a non-router listener."""
+        execution_order: list[str] = []
+
+        class ChainedRouterFlow(Flow):
+            @start()
+            @human_feedback(
+                message="Review:",
+                emit=["approved", "rejected"],
+                llm="gpt-4o-mini",
+            )
+            def draft(self):
+                execution_order.append("draft")
+                return "draft"
+
+            @listen("approved")
+            @human_feedback(
+                message="Final:",
+                emit=["publish", "revise"],
+                llm="gpt-4o-mini",
+            )
+            def final_review(self, prev: HumanFeedbackResult):
+                execution_order.append("final_review")
+                return "final"
+
+            @listen("rejected")
+            def on_rejected(self, prev: HumanFeedbackResult):
+                execution_order.append("on_rejected")
+                return "rejected"
+
+        flow = ChainedRouterFlow()
+
+        with (
+            patch.object(
+                flow, "_request_human_feedback", return_value="bad"
+            ),
+            patch.object(
+                flow, "_collapse_to_outcome", return_value="rejected"
+            ),
+        ):
+            result = flow.kickoff()
+
+        assert execution_order == ["draft", "on_rejected"]
+        assert result == "rejected"
+        assert len(flow.human_feedback_history) == 1
+        assert flow.human_feedback_history[0].outcome == "rejected"
+
+    def test_router_and_non_router_listeners_for_same_outcome(self):
+        """Test that both router and non-router listeners fire for the same outcome."""
+        execution_order: list[str] = []
+
+        class MixedListenerFlow(Flow):
+            @start()
+            @human_feedback(
+                message="Review:",
+                emit=["approved", "rejected"],
+                llm="gpt-4o-mini",
+            )
+            def draft(self):
+                execution_order.append("draft")
+                return "draft"
+
+            @listen("approved")
+            @human_feedback(
+                message="Final:",
+                emit=["publish", "revise"],
+                llm="gpt-4o-mini",
+            )
+            def router_listener(self, prev: HumanFeedbackResult):
+                execution_order.append("router_listener")
+                return "final"
+
+            @listen("approved")
+            def plain_listener(self, prev: HumanFeedbackResult):
+                execution_order.append("plain_listener")
+                return "logged"
+
+            @listen("publish")
+            def on_publish(self, prev: HumanFeedbackResult):
+                execution_order.append("on_publish")
+                return "published"
+
+        flow = MixedListenerFlow()
+
+        with (
+            patch.object(
+                flow,
+                "_request_human_feedback",
+                side_effect=["approve it", "publish it"],
+            ),
+            patch.object(
+                flow,
+                "_collapse_to_outcome",
+                side_effect=["approved", "publish"],
+            ),
+        ):
+            flow.kickoff()
+
+        assert "draft" in execution_order
+        assert "router_listener" in execution_order
+        assert "plain_listener" in execution_order
+        assert "on_publish" in execution_order
+
 
 class TestStateManagement:
     """Tests for state management with human feedback."""