Compare commits

..

13 Commits

Author SHA1 Message Date
Greyson LaLonde
27bd16fd0c fix: use A2UI_EXTENSION_URI for catalog label 2026-03-14 21:28:45 -04:00
Greyson LaLonde
6cc0abbdc6 fix: use typing_extensions.TypedDict for Pydantic compatibility on Python < 3.12 2026-03-14 19:19:21 -04:00
Greyson Lalonde
bc20b96538 refactor: add Field descriptions, explicit defaults, and ConfigDict to A2UI models 2026-03-14 17:25:58 -04:00
Greyson Lalonde
d2e74fc0be refactor: improve typing and use instance state in A2UI client extension 2026-03-14 16:38:05 -04:00
Greyson Lalonde
afb6cbbb6e refactor: add Field descriptions, explicit defaults, and ConfigDict to A2UI catalog models 2026-03-14 15:25:20 -04:00
Greyson Lalonde
e63c6310ed style: consolidate catalog imports in A2UI conformance tests 2026-03-14 13:24:07 -04:00
Greyson LaLonde
aa01d3130f Merge branch 'main' into gl/feat/a2ui-extension 2026-03-14 03:07:44 -04:00
Greyson Lalonde
e6fa652364 fix: replace getattr with direct metadata access on DataPart 2026-03-14 02:55:35 -04:00
Greyson Lalonde
d653a40394 docs: register A2UI guide in docs navigation 2026-03-14 02:48:09 -04:00
Greyson Lalonde
b8f47aaa6b test: add A2UI schema conformance tests 2026-03-14 02:47:09 -04:00
Greyson Lalonde
4ee32fc7d1 feat: add A2UI content type to A2A utils 2026-03-14 02:43:49 -04:00
Greyson Lalonde
c8d776f62a docs: add A2UI extension guide 2026-03-14 02:42:36 -04:00
Greyson Lalonde
b2a2902f00 feat: add A2UI v0.8 extension for declarative UI generation 2026-03-14 02:41:48 -04:00
29 changed files with 6639 additions and 2829 deletions

View File

@@ -351,7 +351,9 @@
"en/learn/using-annotations",
"en/learn/execution-hooks",
"en/learn/llm-hooks",
"en/learn/tool-hooks"
"en/learn/tool-hooks",
"en/learn/a2a-agent-delegation",
"en/learn/a2ui"
]
},
{
@@ -810,7 +812,9 @@
"en/learn/using-annotations",
"en/learn/execution-hooks",
"en/learn/llm-hooks",
"en/learn/tool-hooks"
"en/learn/tool-hooks",
"en/learn/a2a-agent-delegation",
"en/learn/a2ui"
]
},
{

344
docs/en/learn/a2ui.mdx Normal file
View File

@@ -0,0 +1,344 @@
---
title: Agent-to-UI (A2UI) Protocol
description: Enable agents to generate declarative UI surfaces for rich client rendering via the A2UI extension.
icon: window-restore
mode: "wide"
---
## A2UI Overview
A2UI is a declarative UI protocol extension for [A2A](/en/learn/a2a-agent-delegation) that lets agents emit structured JSON messages describing interactive surfaces. Clients receive these messages and render them as rich UI components — forms, cards, lists, modals, and more — without the agent needing to know anything about the client's rendering stack.
A2UI is built on the A2A extension mechanism and identified by the URI `https://a2ui.org/a2a-extension/a2ui/v0.8`.
<Note>
A2UI requires the `a2a-sdk` package. Install with: `uv add 'crewai[a2a]'` or `pip install 'crewai[a2a]'`
</Note>
## How It Works
1. The **server extension** scans agent output for A2UI JSON objects
2. Valid messages are wrapped as `DataPart` entries with the `application/json+a2ui` MIME type
3. The **client extension** augments the agent's system prompt with A2UI instructions and the component catalog
4. The client tracks surface state (active surfaces and data models) across conversation turns
## Server Setup
Add `A2UIServerExtension` to your `A2AServerConfig` to enable A2UI output:
```python Code
from crewai import Agent
from crewai.a2a import A2AServerConfig
from crewai.a2a.extensions.a2ui import A2UIServerExtension
agent = Agent(
role="Dashboard Agent",
goal="Present data through interactive UI surfaces",
backstory="Expert at building clear, actionable dashboards",
llm="gpt-4o",
a2a=A2AServerConfig(
url="https://your-server.com",
extensions=[A2UIServerExtension()],
),
)
```
### Server Extension Options
<ParamField path="catalog_ids" type="list[str] | None" default="None">
Component catalog identifiers the server supports. When set, only these catalogs are advertised to clients.
</ParamField>
<ParamField path="accept_inline_catalogs" type="bool" default="False">
Whether to accept inline catalog definitions from clients in addition to named catalogs.
</ParamField>
## Client Setup
Add `A2UIClientExtension` to your `A2AClientConfig` to enable A2UI rendering:
```python Code
from crewai import Agent
from crewai.a2a import A2AClientConfig
from crewai.a2a.extensions.a2ui import A2UIClientExtension
agent = Agent(
role="UI Coordinator",
goal="Coordinate tasks and render agent responses as rich UI",
backstory="Expert at presenting agent output in interactive formats",
llm="gpt-4o",
a2a=A2AClientConfig(
endpoint="https://dashboard-agent.example.com/.well-known/agent-card.json",
extensions=[A2UIClientExtension()],
),
)
```
### Client Extension Options
<ParamField path="catalog_id" type="str | None" default="None">
Preferred component catalog identifier. Defaults to `"standard (v0.8)"` when not set.
</ParamField>
<ParamField path="allowed_components" type="list[str] | None" default="None">
Restrict which components the agent may use. When `None`, all 18 standard catalog components are available.
</ParamField>
## Message Types
A2UI defines four server-to-client message types. Each message targets a **surface** identified by `surfaceId`.
<Tabs>
<Tab title="beginRendering">
Initializes a new surface with a root component and optional styles.
```json
{
"beginRendering": {
"surfaceId": "dashboard-1",
"root": "main-column",
"catalogId": "standard (v0.8)",
"styles": {
"primaryColor": "#EB6658"
}
}
}
```
</Tab>
<Tab title="surfaceUpdate">
Sends or updates one or more components on an existing surface.
```json
{
"surfaceUpdate": {
"surfaceId": "dashboard-1",
"components": [
{
"id": "main-column",
"component": {
"Column": {
"children": { "explicitList": ["title", "content"] },
"alignment": "start"
}
}
},
{
"id": "title",
"component": {
"Text": {
"text": { "literalString": "Dashboard" },
"usageHint": "h1"
}
}
}
]
}
}
```
</Tab>
<Tab title="dataModelUpdate">
Updates the data model bound to a surface, enabling dynamic content.
```json
{
"dataModelUpdate": {
"surfaceId": "dashboard-1",
"path": "/data/model",
"contents": [
{
"key": "userName",
"valueString": "Alice"
},
{
"key": "score",
"valueNumber": 42
}
]
}
}
```
</Tab>
<Tab title="deleteSurface">
Removes a surface and all its components.
```json
{
"deleteSurface": {
"surfaceId": "dashboard-1"
}
}
```
</Tab>
</Tabs>
## Component Catalog
A2UI ships with 18 standard components organized into three categories:
### Content
| Component | Description | Required Fields |
|-----------|-------------|-----------------|
| **Text** | Renders text with optional heading/body hints | `text` (StringBinding) |
| **Image** | Displays an image with fit and size options | `url` (StringBinding) |
| **Icon** | Renders a named icon from a set of 47 icons | `name` (IconBinding) |
| **Video** | Embeds a video player | `url` (StringBinding) |
| **AudioPlayer** | Embeds an audio player with optional description | `url` (StringBinding) |
### Layout
| Component | Description | Required Fields |
|-----------|-------------|-----------------|
| **Row** | Horizontal flex container | `children` (ChildrenDef) |
| **Column** | Vertical flex container | `children` (ChildrenDef) |
| **List** | Scrollable list (vertical or horizontal) | `children` (ChildrenDef) |
| **Card** | Elevated container for a single child | `child` (str) |
| **Tabs** | Tabbed container | `tabItems` (list of TabItem) |
| **Divider** | Visual separator (horizontal or vertical) | — |
| **Modal** | Overlay triggered by an entry point | `entryPointChild`, `contentChild` (str) |
### Interactive
| Component | Description | Required Fields |
|-----------|-------------|-----------------|
| **Button** | Clickable button that triggers an action | `child` (str), `action` (Action) |
| **CheckBox** | Boolean toggle with a label | `label` (StringBinding), `value` (BooleanBinding) |
| **TextField** | Text input with type and validation options | `label` (StringBinding) |
| **DateTimeInput** | Date and/or time picker | `value` (StringBinding) |
| **MultipleChoice** | Selection from a list of options | `selections` (ArrayBinding), `options` (list) |
| **Slider** | Numeric range slider | `value` (NumberBinding) |
## Data Binding
Components reference values through **bindings** rather than raw literals. This allows surfaces to update dynamically when the data model changes.
There are two ways to bind a value:
- **Literal values** — hardcoded directly in the component definition
- **Path references** — point to a key in the surface's data model
```json
{
"surfaceUpdate": {
"surfaceId": "profile-1",
"components": [
{
"id": "greeting",
"component": {
"Text": {
"text": { "path": "/data/model/userName" },
"usageHint": "h2"
}
}
},
{
"id": "status",
"component": {
"Text": {
"text": { "literalString": "Online" },
"usageHint": "caption"
}
}
}
]
}
}
```
In this example, `greeting` reads the user's name from the data model (updated via `dataModelUpdate`), while `status` uses a hardcoded literal.
## Handling User Actions
Interactive components like `Button` trigger `userAction` events that flow back to the server. Each action includes a `name`, the originating `surfaceId` and `sourceComponentId`, and an optional `context` with key-value pairs.
```json
{
"userAction": {
"name": "submitForm",
"surfaceId": "form-1",
"sourceComponentId": "submit-btn",
"timestamp": "2026-03-12T10:00:00Z",
"context": {
"selectedOption": "optionA"
}
}
}
```
Action context values can also use path bindings to send current data model values back to the server:
```json
{
"Button": {
"child": "confirm-label",
"action": {
"name": "confirm",
"context": [
{
"key": "currentScore",
"value": { "path": "/data/model/score" }
}
]
}
}
}
```
## Validation
Use `validate_a2ui_message` to validate server-to-client messages and `validate_a2ui_event` for client-to-server events:
```python Code
from crewai.a2a.extensions.a2ui import validate_a2ui_message
from crewai.a2a.extensions.a2ui.validator import (
validate_a2ui_event,
A2UIValidationError,
)
# Validate a server message
try:
msg = validate_a2ui_message({"beginRendering": {"surfaceId": "s1", "root": "r1"}})
except A2UIValidationError as exc:
print(exc.errors)
# Validate a client event
try:
event = validate_a2ui_event({
"userAction": {
"name": "click",
"surfaceId": "s1",
"sourceComponentId": "btn-1",
"timestamp": "2026-03-12T10:00:00Z",
}
})
except A2UIValidationError as exc:
print(exc.errors)
```
## Best Practices
<CardGroup cols={2}>
<Card title="Start Simple" icon="play">
Begin with a `beginRendering` message and a single `surfaceUpdate`. Add data binding and interactivity once the basic flow works.
</Card>
<Card title="Use Data Binding for Dynamic Content" icon="arrows-rotate">
Prefer path bindings over literal values for content that changes. Use `dataModelUpdate` to push new values without resending the full component tree.
</Card>
<Card title="Filter Components" icon="filter">
Use the `allowed_components` option on `A2UIClientExtension` to restrict which components the agent may emit, reducing prompt size and keeping output predictable.
</Card>
<Card title="Validate Messages" icon="check">
Use `validate_a2ui_message` and `validate_a2ui_event` to catch malformed payloads early, especially when building custom integrations.
</Card>
</CardGroup>
## Learn More
- [A2A Agent Delegation](/en/learn/a2a-agent-delegation) — configure agents for remote delegation via the A2A protocol
- [A2A Protocol Documentation](https://a2a-protocol.org) — official protocol specification

View File

@@ -1,27 +1,13 @@
# CodeInterpreterTool
## Description
This tool is used to give the Agent the ability to run code (Python3) from the code generated by the Agent itself. The code is executed in a Docker container for secure isolation.
This tool is used to give the Agent the ability to run code (Python3) from the code generated by the Agent itself. The code is executed in a sandboxed environment, so it is safe to run any code.
It is incredibly useful since it allows the Agent to generate code, run it in an isolated environment, get the result and use it to make decisions.
## ⚠️ Security Requirements
**Docker is REQUIRED** for safe code execution. The tool will refuse to execute code without Docker to prevent security vulnerabilities.
### Why Docker is Required
Previous versions included a "restricted sandbox" fallback when Docker was unavailable. This has been **removed** due to critical security vulnerabilities:
- The Python-based sandbox could be escaped via object introspection
- Attackers could recover the original `__import__` function and access any module
- This allowed arbitrary command execution on the host system
**Docker provides real process isolation** and is the only secure way to execute untrusted code.
It is incredible useful since it allows the Agent to generate code, run it in the same environment, get the result and use it to make decisions.
## Requirements
- **Docker (REQUIRED)** - Install from [docker.com](https://docs.docker.com/get-docker/)
- Docker
## Installation
Install the crewai_tools package
@@ -31,9 +17,7 @@ pip install 'crewai[tools]'
## Example
Remember that when using this tool, the code must be generated by the Agent itself. The code must be Python3 code. It will take some time the first time to run because it needs to build the Docker image.
### Basic Usage (Docker Container - Recommended)
Remember that when using this tool, the code must be generated by the Agent itself. The code must be a Python3 code. And it will take some time for the first time to run because it needs to build the Docker image.
```python
from crewai_tools import CodeInterpreterTool
@@ -44,9 +28,7 @@ Agent(
)
```
### Custom Dockerfile
If you need to pass your own Dockerfile:
Or if you need to pass your own Dockerfile just do this
```python
from crewai_tools import CodeInterpreterTool
@@ -57,39 +39,15 @@ Agent(
)
```
### Manual Docker Host Configuration
If it is difficult to connect to the Docker daemon automatically (especially for macOS users), you can set up the Docker host manually:
If it is difficult to connect to docker daemon automatically (especially for macOS users), you can do this to setup docker host manually
```python
from crewai_tools import CodeInterpreterTool
Agent(
...
tools=[CodeInterpreterTool(
user_docker_base_url="<Docker Host Base Url>",
user_dockerfile_path="<Dockerfile_path>"
)],
tools=[CodeInterpreterTool(user_docker_base_url="<Docker Host Base Url>",
user_dockerfile_path="<Dockerfile_path>")],
)
```
### Unsafe Mode (NOT RECOMMENDED)
If you absolutely cannot use Docker and **fully trust the code source**, you can use unsafe mode:
```python
from crewai_tools import CodeInterpreterTool
# WARNING: Only use with fully trusted code!
Agent(
...
tools=[CodeInterpreterTool(unsafe_mode=True)],
)
```
**⚠️ SECURITY WARNING:** `unsafe_mode=True` executes code directly on the host without any isolation. Only use this if:
- You completely trust the code being executed
- You understand the security risks
- You cannot install Docker in your environment
For production use, **always use Docker** (the default mode).

View File

@@ -50,16 +50,11 @@ class CodeInterpreterSchema(BaseModel):
class SandboxPython:
"""INSECURE: A restricted Python execution environment with known vulnerabilities.
"""A restricted Python execution environment for running code safely.
WARNING: This class does NOT provide real security isolation and is vulnerable to
sandbox escape attacks via Python object introspection. Attackers can recover the
original __import__ function and bypass all restrictions.
DO NOT USE for untrusted code execution. Use Docker containers instead.
This class attempts to restrict access to dangerous modules and built-in functions
but provides no real security boundary against a motivated attacker.
This class provides methods to safely execute Python code by restricting access to
potentially dangerous modules and built-in functions. It creates a sandboxed
environment where harmful operations are blocked.
"""
BLOCKED_MODULES: ClassVar[set[str]] = {
@@ -304,8 +299,8 @@ class CodeInterpreterTool(BaseTool):
def run_code_safety(self, code: str, libraries_used: list[str]) -> str:
"""Runs code in the safest available environment.
Requires Docker to be available for secure code execution. Fails closed
if Docker is not available to prevent sandbox escape vulnerabilities.
Attempts to run code in Docker if available, falls back to a restricted
sandbox if Docker is not available.
Args:
code: The Python code to execute as a string.
@@ -313,24 +308,10 @@ class CodeInterpreterTool(BaseTool):
Returns:
The output of the executed code as a string.
Raises:
RuntimeError: If Docker is not available, as the restricted sandbox
is vulnerable to escape attacks and should not be used
for untrusted code execution.
"""
if self._check_docker_available():
return self.run_code_in_docker(code, libraries_used)
error_msg = (
"Docker is required for safe code execution but is not available. "
"The restricted sandbox fallback has been removed due to security vulnerabilities "
"that allow sandbox escape via Python object introspection. "
"Please install Docker (https://docs.docker.com/get-docker/) or use unsafe_mode=True "
"if you trust the code source and understand the security risks."
)
Printer.print(error_msg, color="bold_red")
raise RuntimeError(error_msg)
return self.run_code_in_restricted_sandbox(code)
def run_code_in_docker(self, code: str, libraries_used: list[str]) -> str:
"""Runs Python code in a Docker container for safe isolation.
@@ -361,19 +342,10 @@ class CodeInterpreterTool(BaseTool):
@staticmethod
def run_code_in_restricted_sandbox(code: str) -> str:
"""DEPRECATED AND INSECURE: Runs Python code in a restricted sandbox environment.
"""Runs Python code in a restricted sandbox environment.
WARNING: This method is vulnerable to sandbox escape attacks via Python object
introspection and should NOT be used for untrusted code execution. It has been
deprecated and is only kept for backward compatibility with trusted code.
The "restricted" environment can be bypassed by attackers who can:
- Use object graph introspection to recover the original __import__ function
- Access any Python module including os, subprocess, sys, etc.
- Execute arbitrary commands on the host system
Use run_code_in_docker() for secure code execution, or run_code_unsafe()
if you explicitly acknowledge the security risks.
Executes the code with restricted access to potentially dangerous modules and
built-in functions for basic safety when Docker is not available.
Args:
code: The Python code to execute as a string.
@@ -382,10 +354,7 @@ class CodeInterpreterTool(BaseTool):
The value of the 'result' variable from the executed code,
or an error message if execution failed.
"""
Printer.print(
"WARNING: Running code in INSECURE restricted sandbox (vulnerable to escape attacks)",
color="bold_red"
)
Printer.print("Running code in restricted sandbox", color="yellow")
exec_locals: dict[str, Any] = {}
try:
SandboxPython.exec(code=code, locals_=exec_locals)

View File

@@ -76,22 +76,24 @@ print("This is line 2")"""
)
def test_docker_unavailable_raises_error(printer_mock, docker_unavailable_mock):
"""Test that execution fails when Docker is unavailable in safe mode."""
def test_restricted_sandbox_basic_code_execution(printer_mock, docker_unavailable_mock):
"""Test basic code execution."""
tool = CodeInterpreterTool()
code = """
result = 2 + 2
print(result)
"""
with pytest.raises(RuntimeError) as exc_info:
tool.run(code=code, libraries_used=[])
assert "Docker is required for safe code execution" in str(exc_info.value)
assert "sandbox escape" in str(exc_info.value)
result = tool.run(code=code, libraries_used=[])
printer_mock.assert_called_with(
"Running code in restricted sandbox", color="yellow"
)
assert result == 4
def test_restricted_sandbox_running_with_blocked_modules():
"""Test that restricted modules cannot be imported when using the deprecated sandbox directly."""
def test_restricted_sandbox_running_with_blocked_modules(
printer_mock, docker_unavailable_mock
):
"""Test that restricted modules cannot be imported."""
tool = CodeInterpreterTool()
restricted_modules = SandboxPython.BLOCKED_MODULES
@@ -100,15 +102,18 @@ def test_restricted_sandbox_running_with_blocked_modules():
import {module}
result = "Import succeeded"
"""
# Note: run_code_in_restricted_sandbox is deprecated and insecure
# This test verifies the old behavior but should not be used in production
result = tool.run_code_in_restricted_sandbox(code)
result = tool.run(code=code, libraries_used=[])
printer_mock.assert_called_with(
"Running code in restricted sandbox", color="yellow"
)
assert f"An error occurred: Importing '{module}' is not allowed" in result
def test_restricted_sandbox_running_with_blocked_builtins():
"""Test that restricted builtins are not available when using the deprecated sandbox directly."""
def test_restricted_sandbox_running_with_blocked_builtins(
printer_mock, docker_unavailable_mock
):
"""Test that restricted builtins are not available."""
tool = CodeInterpreterTool()
restricted_builtins = SandboxPython.UNSAFE_BUILTINS
@@ -117,23 +122,25 @@ def test_restricted_sandbox_running_with_blocked_builtins():
{builtin}("test")
result = "Builtin available"
"""
# Note: run_code_in_restricted_sandbox is deprecated and insecure
# This test verifies the old behavior but should not be used in production
result = tool.run_code_in_restricted_sandbox(code)
result = tool.run(code=code, libraries_used=[])
printer_mock.assert_called_with(
"Running code in restricted sandbox", color="yellow"
)
assert f"An error occurred: name '{builtin}' is not defined" in result
def test_restricted_sandbox_running_with_no_result_variable(
printer_mock, docker_unavailable_mock
):
"""Test behavior when no result variable is set in deprecated sandbox."""
"""Test behavior when no result variable is set."""
tool = CodeInterpreterTool()
code = """
x = 10
"""
# Note: run_code_in_restricted_sandbox is deprecated and insecure
# This test verifies the old behavior but should not be used in production
result = tool.run_code_in_restricted_sandbox(code)
result = tool.run(code=code, libraries_used=[])
printer_mock.assert_called_with(
"Running code in restricted sandbox", color="yellow"
)
assert result == "No result variable found."
@@ -165,50 +172,3 @@ result = eval("5/1")
"WARNING: Running code in unsafe mode", color="bold_magenta"
)
assert 5.0 == result
@pytest.mark.xfail(
reason=(
"run_code_in_restricted_sandbox is known to be vulnerable to sandbox "
"escape via object introspection. This test encodes the desired secure "
"behavior (no escape possible) and will start passing once the "
"vulnerability is fixed or the function is removed."
)
)
def test_sandbox_escape_vulnerability_demonstration(printer_mock):
"""Demonstrate that the restricted sandbox is vulnerable to escape attacks.
This test shows that an attacker can use Python object introspection to bypass
the restricted sandbox and access blocked modules like 'os'. This is why the
sandbox should never be used for untrusted code execution.
NOTE: This test uses the deprecated run_code_in_restricted_sandbox directly
to demonstrate the vulnerability. In production, Docker is now required.
"""
tool = CodeInterpreterTool()
# Classic Python sandbox escape via object introspection
escape_code = """
# Recover the real __import__ function via object introspection
for cls in ().__class__.__bases__[0].__subclasses__():
if cls.__name__ == 'catch_warnings':
# Get the real builtins module
real_builtins = cls()._module.__builtins__
real_import = real_builtins['__import__']
# Now we can import os and execute commands
os = real_import('os')
# Demonstrate we have escaped the sandbox
result = "SANDBOX_ESCAPED" if hasattr(os, 'system') else "FAILED"
break
"""
# The deprecated sandbox is vulnerable to this attack
result = tool.run_code_in_restricted_sandbox(escape_code)
# Desired behavior: the restricted sandbox should prevent this escape.
# If this assertion fails, run_code_in_restricted_sandbox remains vulnerable.
assert result != "SANDBOX_ESCAPED", (
"The restricted sandbox was bypassed via object introspection. "
"This indicates run_code_in_restricted_sandbox is still vulnerable and "
"is why Docker is now required for safe code execution."
)

View File

@@ -0,0 +1,68 @@
"""A2UI (Agent to UI) declarative UI protocol support for CrewAI."""
from crewai.a2a.extensions.a2ui.catalog import (
AudioPlayer,
Button,
Card,
CheckBox,
Column,
DateTimeInput,
Divider,
Icon,
Image,
List,
Modal,
MultipleChoice,
Row,
Slider,
Tabs,
Text,
TextField,
Video,
)
from crewai.a2a.extensions.a2ui.client_extension import A2UIClientExtension
from crewai.a2a.extensions.a2ui.models import (
A2UIEvent,
A2UIMessage,
A2UIResponse,
BeginRendering,
DataModelUpdate,
DeleteSurface,
SurfaceUpdate,
UserAction,
)
from crewai.a2a.extensions.a2ui.server_extension import A2UIServerExtension
from crewai.a2a.extensions.a2ui.validator import validate_a2ui_message
__all__ = [
"A2UIClientExtension",
"A2UIEvent",
"A2UIMessage",
"A2UIResponse",
"A2UIServerExtension",
"AudioPlayer",
"BeginRendering",
"Button",
"Card",
"CheckBox",
"Column",
"DataModelUpdate",
"DateTimeInput",
"DeleteSurface",
"Divider",
"Icon",
"Image",
"List",
"Modal",
"MultipleChoice",
"Row",
"Slider",
"SurfaceUpdate",
"Tabs",
"Text",
"TextField",
"UserAction",
"Video",
"validate_a2ui_message",
]

View File

@@ -0,0 +1,467 @@
"""Typed helpers for A2UI standard catalog components.
These models provide optional type safety for standard catalog components.
Agents can also use raw dicts validated against the JSON schema.
"""
from __future__ import annotations
from typing import Literal
from pydantic import BaseModel, ConfigDict, Field
class StringBinding(BaseModel):
"""A string value: literal or data-model path."""
literal_string: str | None = Field(
default=None, alias="literalString", description="Literal string value."
)
path: str | None = Field(default=None, description="Data-model path reference.")
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class NumberBinding(BaseModel):
"""A numeric value: literal or data-model path."""
literal_number: float | None = Field(
default=None, alias="literalNumber", description="Literal numeric value."
)
path: str | None = Field(default=None, description="Data-model path reference.")
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class BooleanBinding(BaseModel):
"""A boolean value: literal or data-model path."""
literal_boolean: bool | None = Field(
default=None, alias="literalBoolean", description="Literal boolean value."
)
path: str | None = Field(default=None, description="Data-model path reference.")
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class ArrayBinding(BaseModel):
"""An array value: literal or data-model path."""
literal_array: list[str] | None = Field(
default=None, alias="literalArray", description="Literal array of strings."
)
path: str | None = Field(default=None, description="Data-model path reference.")
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class ChildrenDef(BaseModel):
"""Children definition for layout components."""
explicit_list: list[str] | None = Field(
default=None,
alias="explicitList",
description="Explicit list of child component IDs.",
)
template: ChildTemplate | None = Field(
default=None, description="Template for generating dynamic children."
)
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class ChildTemplate(BaseModel):
"""Template for generating dynamic children from a data model list."""
component_id: str = Field(
alias="componentId", description="ID of the component to repeat."
)
data_binding: str = Field(
alias="dataBinding", description="Data-model path to bind the template to."
)
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class ActionContextEntry(BaseModel):
"""A key-value pair in an action context payload."""
key: str = Field(description="Context entry key.")
value: ActionBoundValue = Field(description="Context entry value.")
model_config = ConfigDict(extra="forbid")
class ActionBoundValue(BaseModel):
"""A value in an action context: literal or data-model path."""
path: str | None = Field(default=None, description="Data-model path reference.")
literal_string: str | None = Field(
default=None, alias="literalString", description="Literal string value."
)
literal_number: float | None = Field(
default=None, alias="literalNumber", description="Literal numeric value."
)
literal_boolean: bool | None = Field(
default=None, alias="literalBoolean", description="Literal boolean value."
)
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class Action(BaseModel):
"""Client-side action dispatched by interactive components."""
name: str = Field(description="Action name dispatched on interaction.")
context: list[ActionContextEntry] | None = Field(
default=None, description="Key-value pairs sent with the action."
)
model_config = ConfigDict(extra="forbid")
class TabItem(BaseModel):
"""A single tab definition."""
title: StringBinding = Field(description="Tab title text.")
child: str = Field(description="Component ID rendered as the tab content.")
model_config = ConfigDict(extra="forbid")
class MultipleChoiceOption(BaseModel):
"""A single option in a MultipleChoice component."""
label: StringBinding = Field(description="Display label for the option.")
value: str = Field(description="Value submitted when the option is selected.")
model_config = ConfigDict(extra="forbid")
class Text(BaseModel):
"""Displays text content."""
text: StringBinding = Field(description="Text content to display.")
usage_hint: Literal["h1", "h2", "h3", "h4", "h5", "caption", "body"] | None = Field(
default=None, alias="usageHint", description="Semantic hint for text styling."
)
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class Image(BaseModel):
"""Displays an image."""
url: StringBinding = Field(description="Image source URL.")
fit: Literal["contain", "cover", "fill", "none", "scale-down"] | None = Field(
default=None, description="Object-fit behavior for the image."
)
usage_hint: (
Literal[
"icon", "avatar", "smallFeature", "mediumFeature", "largeFeature", "header"
]
| None
) = Field(
default=None, alias="usageHint", description="Semantic hint for image sizing."
)
model_config = ConfigDict(populate_by_name=True, extra="forbid")
IconName = Literal[
"accountCircle",
"add",
"arrowBack",
"arrowForward",
"attachFile",
"calendarToday",
"call",
"camera",
"check",
"close",
"delete",
"download",
"edit",
"event",
"error",
"favorite",
"favoriteOff",
"folder",
"help",
"home",
"info",
"locationOn",
"lock",
"lockOpen",
"mail",
"menu",
"moreVert",
"moreHoriz",
"notificationsOff",
"notifications",
"payment",
"person",
"phone",
"photo",
"print",
"refresh",
"search",
"send",
"settings",
"share",
"shoppingCart",
"star",
"starHalf",
"starOff",
"upload",
"visibility",
"visibilityOff",
"warning",
]
class IconBinding(BaseModel):
"""Icon name: literal enum or data-model path."""
literal_string: IconName | None = Field(
default=None, alias="literalString", description="Literal icon name."
)
path: str | None = Field(default=None, description="Data-model path reference.")
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class Icon(BaseModel):
"""Displays a named icon."""
name: IconBinding = Field(description="Icon name binding.")
model_config = ConfigDict(extra="forbid")
class Video(BaseModel):
"""Displays a video player."""
url: StringBinding = Field(description="Video source URL.")
model_config = ConfigDict(extra="forbid")
class AudioPlayer(BaseModel):
"""Displays an audio player."""
url: StringBinding = Field(description="Audio source URL.")
description: StringBinding | None = Field(
default=None, description="Accessible description of the audio content."
)
model_config = ConfigDict(extra="forbid")
class Row(BaseModel):
"""Horizontal layout container."""
children: ChildrenDef = Field(description="Child components in this row.")
distribution: (
Literal["center", "end", "spaceAround", "spaceBetween", "spaceEvenly", "start"]
| None
) = Field(
default=None, description="How children are distributed along the main axis."
)
alignment: Literal["start", "center", "end", "stretch"] | None = Field(
default=None, description="How children are aligned on the cross axis."
)
model_config = ConfigDict(extra="forbid")
class Column(BaseModel):
"""Vertical layout container."""
children: ChildrenDef = Field(description="Child components in this column.")
distribution: (
Literal["start", "center", "end", "spaceBetween", "spaceAround", "spaceEvenly"]
| None
) = Field(
default=None, description="How children are distributed along the main axis."
)
alignment: Literal["center", "end", "start", "stretch"] | None = Field(
default=None, description="How children are aligned on the cross axis."
)
model_config = ConfigDict(extra="forbid")
class List(BaseModel):
"""Scrollable list container."""
children: ChildrenDef = Field(description="Child components in this list.")
direction: Literal["vertical", "horizontal"] | None = Field(
default=None, description="Scroll direction of the list."
)
alignment: Literal["start", "center", "end", "stretch"] | None = Field(
default=None, description="How children are aligned on the cross axis."
)
model_config = ConfigDict(extra="forbid")
class Card(BaseModel):
"""Card container wrapping a single child."""
child: str = Field(description="Component ID of the card content.")
model_config = ConfigDict(extra="forbid")
class Tabs(BaseModel):
"""Tabbed navigation container."""
tab_items: list[TabItem] = Field(
alias="tabItems", description="List of tab definitions."
)
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class Divider(BaseModel):
"""A visual divider line."""
axis: Literal["horizontal", "vertical"] | None = Field(
default=None, description="Orientation of the divider."
)
model_config = ConfigDict(extra="forbid")
class Modal(BaseModel):
"""A modal dialog with an entry point trigger and content."""
entry_point_child: str = Field(
alias="entryPointChild", description="Component ID that triggers the modal."
)
content_child: str = Field(
alias="contentChild", description="Component ID rendered inside the modal."
)
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class Button(BaseModel):
"""An interactive button with an action."""
child: str = Field(description="Component ID of the button label.")
primary: bool | None = Field(
default=None, description="Whether the button uses primary styling."
)
action: Action = Field(description="Action dispatched when the button is clicked.")
model_config = ConfigDict(extra="forbid")
class CheckBox(BaseModel):
"""A checkbox input."""
label: StringBinding = Field(description="Label text for the checkbox.")
value: BooleanBinding = Field(
description="Boolean value binding for the checkbox state."
)
model_config = ConfigDict(extra="forbid")
class TextField(BaseModel):
"""A text input field."""
label: StringBinding = Field(description="Label text for the input.")
text: StringBinding | None = Field(
default=None, description="Current text value binding."
)
text_field_type: (
Literal["date", "longText", "number", "shortText", "obscured"] | None
) = Field(default=None, alias="textFieldType", description="Input type variant.")
validation_regexp: str | None = Field(
default=None,
alias="validationRegexp",
description="Regex pattern for client-side validation.",
)
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class DateTimeInput(BaseModel):
"""A date and/or time picker."""
value: StringBinding = Field(description="ISO date/time string value binding.")
enable_date: bool | None = Field(
default=None,
alias="enableDate",
description="Whether the date picker is enabled.",
)
enable_time: bool | None = Field(
default=None,
alias="enableTime",
description="Whether the time picker is enabled.",
)
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class MultipleChoice(BaseModel):
"""A multiple-choice selection component."""
selections: ArrayBinding = Field(description="Array binding for selected values.")
options: list[MultipleChoiceOption] = Field(description="Available choices.")
max_allowed_selections: int | None = Field(
default=None,
alias="maxAllowedSelections",
description="Maximum number of selections allowed.",
)
variant: Literal["checkbox", "chips"] | None = Field(
default=None, description="Visual variant for the selection UI."
)
filterable: bool | None = Field(
default=None, description="Whether options can be filtered by typing."
)
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class Slider(BaseModel):
"""A numeric slider input."""
value: NumberBinding = Field(
description="Numeric value binding for the slider position."
)
min_value: float | None = Field(
default=None, alias="minValue", description="Minimum slider value."
)
max_value: float | None = Field(
default=None, alias="maxValue", description="Maximum slider value."
)
model_config = ConfigDict(populate_by_name=True, extra="forbid")
STANDARD_CATALOG_COMPONENTS: frozenset[str] = frozenset(
{
"Text",
"Image",
"Icon",
"Video",
"AudioPlayer",
"Row",
"Column",
"List",
"Card",
"Tabs",
"Divider",
"Modal",
"Button",
"CheckBox",
"TextField",
"DateTimeInput",
"MultipleChoice",
"Slider",
}
)

View File

@@ -0,0 +1,285 @@
"""A2UI client extension for the A2A protocol."""
from __future__ import annotations
from collections.abc import Sequence
import logging
from typing import TYPE_CHECKING, Any, cast
from pydantic import Field
from pydantic.dataclasses import dataclass
from typing_extensions import TypedDict
from crewai.a2a.extensions.a2ui.models import extract_a2ui_json_objects
from crewai.a2a.extensions.a2ui.prompt import build_a2ui_system_prompt
from crewai.a2a.extensions.a2ui.server_extension import A2UI_MIME_TYPE
from crewai.a2a.extensions.a2ui.validator import (
A2UIValidationError,
validate_a2ui_message,
)
if TYPE_CHECKING:
from a2a.types import Message
from crewai.agent.core import Agent
logger = logging.getLogger(__name__)
class StylesDict(TypedDict, total=False):
"""Serialized surface styling."""
font: str
primaryColor: str
class ComponentEntryDict(TypedDict, total=False):
"""Serialized component entry in a surface update."""
id: str
weight: float
component: dict[str, Any]
class BeginRenderingDict(TypedDict, total=False):
"""Serialized beginRendering payload."""
surfaceId: str
root: str
catalogId: str
styles: StylesDict
class SurfaceUpdateDict(TypedDict, total=False):
"""Serialized surfaceUpdate payload."""
surfaceId: str
components: list[ComponentEntryDict]
class DataEntryDict(TypedDict, total=False):
"""Serialized data model entry."""
key: str
valueString: str
valueNumber: float
valueBoolean: bool
valueMap: list[DataEntryDict]
class DataModelUpdateDict(TypedDict, total=False):
"""Serialized dataModelUpdate payload."""
surfaceId: str
path: str
contents: list[DataEntryDict]
class DeleteSurfaceDict(TypedDict):
"""Serialized deleteSurface payload."""
surfaceId: str
class A2UIMessageDict(TypedDict, total=False):
"""Serialized A2UI server-to-client message with exactly one key set."""
beginRendering: BeginRenderingDict
surfaceUpdate: SurfaceUpdateDict
dataModelUpdate: DataModelUpdateDict
deleteSurface: DeleteSurfaceDict
@dataclass
class A2UIConversationState:
"""Tracks active A2UI surfaces and data models across a conversation."""
active_surfaces: dict[str, dict[str, Any]] = Field(default_factory=dict)
data_models: dict[str, list[dict[str, Any]]] = Field(default_factory=dict)
last_a2ui_messages: list[A2UIMessageDict] = Field(default_factory=list)
def is_ready(self) -> bool:
"""Return True when at least one surface is active."""
return bool(self.active_surfaces)
class A2UIClientExtension:
"""A2A client extension that adds A2UI support to agents.
Implements the ``A2AExtension`` protocol to inject A2UI prompt
instructions, track UI state across conversations, and validate
A2UI messages in responses.
Example::
A2AClientConfig(
endpoint="...",
extensions=["https://a2ui.org/a2a-extension/a2ui/v0.8"],
client_extensions=[A2UIClientExtension()],
)
"""
def __init__(
self,
catalog_id: str | None = None,
allowed_components: list[str] | None = None,
) -> None:
"""Initialize the A2UI client extension.
Args:
catalog_id: Catalog identifier to use for prompt generation.
allowed_components: Subset of component names to expose to the agent.
"""
self._catalog_id = catalog_id
self._allowed_components = allowed_components
def inject_tools(self, agent: Agent) -> None:
"""No-op — A2UI uses prompt augmentation rather than tool injection."""
def extract_state_from_history(
self, conversation_history: Sequence[Message]
) -> A2UIConversationState | None:
"""Scan conversation history for A2UI DataParts and track surface state.
When ``catalog_id`` is set, only surfaces matching that catalog are tracked.
"""
state = A2UIConversationState()
for message in conversation_history:
for part in message.parts:
root = part.root
if root.kind != "data":
continue
metadata = root.metadata or {}
mime_type = metadata.get("mimeType", "")
if mime_type != A2UI_MIME_TYPE:
continue
data = root.data
if not isinstance(data, dict):
continue
surface_id = _get_surface_id(data)
if not surface_id:
continue
if self._catalog_id and "beginRendering" in data:
catalog_id = data["beginRendering"].get("catalogId")
if catalog_id and catalog_id != self._catalog_id:
continue
if "deleteSurface" in data:
state.active_surfaces.pop(surface_id, None)
state.data_models.pop(surface_id, None)
elif "beginRendering" in data:
state.active_surfaces[surface_id] = data["beginRendering"]
elif "surfaceUpdate" in data:
state.active_surfaces[surface_id] = data["surfaceUpdate"]
elif "dataModelUpdate" in data:
contents = data["dataModelUpdate"].get("contents", [])
state.data_models.setdefault(surface_id, []).extend(contents)
if not state.active_surfaces and not state.data_models:
return None
return state
def augment_prompt(
self,
base_prompt: str,
_conversation_state: A2UIConversationState | None,
) -> str:
"""Append A2UI system prompt instructions to the base prompt."""
a2ui_prompt = build_a2ui_system_prompt(
catalog_id=self._catalog_id,
allowed_components=self._allowed_components,
)
return f"{base_prompt}\n\n{a2ui_prompt}"
def process_response(
self,
agent_response: Any,
conversation_state: A2UIConversationState | None,
) -> Any:
"""Extract and validate A2UI JSON from agent output.
When ``allowed_components`` is set, components not in the allowlist are
logged and stripped from surface updates. Stores extracted A2UI messages
on the conversation state and returns the original response unchanged.
"""
text = (
agent_response if isinstance(agent_response, str) else str(agent_response)
)
a2ui_messages = _extract_and_validate(text)
if self._allowed_components:
allowed = set(self._allowed_components)
a2ui_messages = [_filter_components(msg, allowed) for msg in a2ui_messages]
if a2ui_messages and conversation_state is not None:
conversation_state.last_a2ui_messages = a2ui_messages
return agent_response
def _get_surface_id(data: dict[str, Any]) -> str | None:
"""Extract surfaceId from any A2UI message type."""
for key in ("beginRendering", "surfaceUpdate", "dataModelUpdate", "deleteSurface"):
inner = data.get(key)
if isinstance(inner, dict):
sid = inner.get("surfaceId")
if isinstance(sid, str):
return sid
return None
def _filter_components(msg: A2UIMessageDict, allowed: set[str]) -> A2UIMessageDict:
"""Strip components whose type is not in *allowed* from a surfaceUpdate."""
surface_update = msg.get("surfaceUpdate")
if not isinstance(surface_update, dict):
return msg
components = surface_update.get("components")
if not isinstance(components, list):
return msg
filtered = []
for entry in components:
component = entry.get("component", {})
component_types = set(component.keys())
disallowed = component_types - allowed
if disallowed:
logger.debug(
"Stripping disallowed component type(s) %s from surface update",
disallowed,
)
continue
filtered.append(entry)
if len(filtered) == len(components):
return msg
return {**msg, "surfaceUpdate": {**surface_update, "components": filtered}}
def _extract_and_validate(text: str) -> list[A2UIMessageDict]:
"""Extract A2UI JSON objects from text and validate them."""
return [
dumped
for candidate in extract_a2ui_json_objects(text)
if (dumped := _try_validate(candidate)) is not None
]
def _try_validate(candidate: dict[str, Any]) -> A2UIMessageDict | None:
"""Validate a single A2UI candidate, returning None on failure."""
try:
msg = validate_a2ui_message(candidate)
except A2UIValidationError:
logger.debug(
"Skipping invalid A2UI candidate in agent output",
exc_info=True,
)
return None
return cast(A2UIMessageDict, msg.model_dump(by_alias=True, exclude_none=True))

View File

@@ -0,0 +1,281 @@
"""Pydantic models for A2UI server-to-client messages and client-to-server events."""
from __future__ import annotations
import json
import re
from typing import Any
from pydantic import BaseModel, ConfigDict, Field, model_validator
class BoundValue(BaseModel):
"""A value that can be a literal or a data-model path reference."""
literal_string: str | None = Field(
default=None, alias="literalString", description="Literal string value."
)
literal_number: float | None = Field(
default=None, alias="literalNumber", description="Literal numeric value."
)
literal_boolean: bool | None = Field(
default=None, alias="literalBoolean", description="Literal boolean value."
)
literal_array: list[str] | None = Field(
default=None, alias="literalArray", description="Literal array of strings."
)
path: str | None = Field(default=None, description="Data-model path reference.")
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class MapEntry(BaseModel):
"""A single entry in a valueMap adjacency list, supporting recursive nesting."""
key: str = Field(description="Entry key.")
value_string: str | None = Field(
default=None, alias="valueString", description="String value."
)
value_number: float | None = Field(
default=None, alias="valueNumber", description="Numeric value."
)
value_boolean: bool | None = Field(
default=None, alias="valueBoolean", description="Boolean value."
)
value_map: list[MapEntry] | None = Field(
default=None, alias="valueMap", description="Nested map entries."
)
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class DataEntry(BaseModel):
"""A data model entry with a key and exactly one typed value."""
key: str = Field(description="Entry key.")
value_string: str | None = Field(
default=None, alias="valueString", description="String value."
)
value_number: float | None = Field(
default=None, alias="valueNumber", description="Numeric value."
)
value_boolean: bool | None = Field(
default=None, alias="valueBoolean", description="Boolean value."
)
value_map: list[MapEntry] | None = Field(
default=None, alias="valueMap", description="Nested map entries."
)
model_config = ConfigDict(populate_by_name=True, extra="forbid")
_HEX_COLOR_PATTERN: re.Pattern[str] = re.compile(r"^#[0-9a-fA-F]{6}$")
class Styles(BaseModel):
"""Surface styling information."""
font: str | None = Field(default=None, description="Font family name.")
primary_color: str | None = Field(
default=None,
alias="primaryColor",
pattern=_HEX_COLOR_PATTERN.pattern,
description="Primary color as a hex string.",
)
model_config = ConfigDict(populate_by_name=True, extra="allow")
class ComponentEntry(BaseModel):
"""A single component in a UI widget tree.
The ``component`` dict must contain exactly one key — the component type
name (e.g. ``"Text"``, ``"Button"``) — whose value holds the component
properties. Component internals are left as ``dict[str, Any]`` because
they are catalog-dependent; use the typed helpers in ``catalog.py`` for
the standard catalog.
"""
id: str = Field(description="Unique component identifier.")
weight: float | None = Field(
default=None, description="Flex weight for layout distribution."
)
component: dict[str, Any] = Field(
description="Component type name mapped to its properties."
)
model_config = ConfigDict(extra="forbid")
class BeginRendering(BaseModel):
"""Signals the client to begin rendering a surface."""
surface_id: str = Field(alias="surfaceId", description="Unique surface identifier.")
root: str = Field(description="Component ID of the root element.")
catalog_id: str | None = Field(
default=None,
alias="catalogId",
description="Catalog identifier for the surface.",
)
styles: Styles | None = Field(
default=None, description="Surface styling overrides."
)
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class SurfaceUpdate(BaseModel):
"""Updates a surface with a new set of components."""
surface_id: str = Field(alias="surfaceId", description="Target surface identifier.")
components: list[ComponentEntry] = Field(
min_length=1, description="Components to render on the surface."
)
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class DataModelUpdate(BaseModel):
"""Updates the data model for a surface."""
surface_id: str = Field(alias="surfaceId", description="Target surface identifier.")
path: str | None = Field(
default=None, description="Data-model path prefix for the update."
)
contents: list[DataEntry] = Field(
description="Data entries to merge into the model."
)
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class DeleteSurface(BaseModel):
"""Signals the client to delete a surface."""
surface_id: str = Field(
alias="surfaceId", description="Surface identifier to delete."
)
model_config = ConfigDict(populate_by_name=True, extra="forbid")
class A2UIMessage(BaseModel):
"""Union wrapper for the four server-to-client A2UI message types.
Exactly one of the fields must be set.
"""
begin_rendering: BeginRendering | None = Field(
default=None,
alias="beginRendering",
description="Begin rendering a new surface.",
)
surface_update: SurfaceUpdate | None = Field(
default=None,
alias="surfaceUpdate",
description="Update components on a surface.",
)
data_model_update: DataModelUpdate | None = Field(
default=None,
alias="dataModelUpdate",
description="Update the surface data model.",
)
delete_surface: DeleteSurface | None = Field(
default=None, alias="deleteSurface", description="Delete an existing surface."
)
model_config = ConfigDict(populate_by_name=True, extra="forbid")
@model_validator(mode="after")
def _check_exactly_one(self) -> A2UIMessage:
"""Enforce the spec's exactly-one-of constraint."""
fields = [
self.begin_rendering,
self.surface_update,
self.data_model_update,
self.delete_surface,
]
count = sum(f is not None for f in fields)
if count != 1:
raise ValueError(f"Exactly one A2UI message type must be set, got {count}")
return self
class UserAction(BaseModel):
"""Reports a user-initiated action from a component."""
name: str = Field(description="Action name.")
surface_id: str = Field(alias="surfaceId", description="Source surface identifier.")
source_component_id: str = Field(
alias="sourceComponentId", description="Component that triggered the action."
)
timestamp: str = Field(description="ISO 8601 timestamp of the action.")
context: dict[str, Any] = Field(description="Action context payload.")
model_config = ConfigDict(populate_by_name=True)
class ClientError(BaseModel):
"""Reports a client-side error."""
model_config = ConfigDict(extra="allow")
class A2UIEvent(BaseModel):
"""Union wrapper for client-to-server events."""
user_action: UserAction | None = Field(
default=None, alias="userAction", description="User-initiated action event."
)
error: ClientError | None = Field(
default=None, description="Client-side error report."
)
model_config = ConfigDict(populate_by_name=True)
@model_validator(mode="after")
def _check_exactly_one(self) -> A2UIEvent:
"""Enforce the spec's exactly-one-of constraint."""
fields = [self.user_action, self.error]
count = sum(f is not None for f in fields)
if count != 1:
raise ValueError(f"Exactly one A2UI event type must be set, got {count}")
return self
class A2UIResponse(BaseModel):
"""Typed wrapper for responses containing A2UI messages."""
text: str = Field(description="Raw text content of the response.")
a2ui_parts: list[dict[str, Any]] = Field(
default_factory=list, description="A2UI DataParts extracted from the response."
)
a2ui_messages: list[dict[str, Any]] = Field(
default_factory=list, description="Validated A2UI message dicts."
)
_A2UI_KEYS = {"beginRendering", "surfaceUpdate", "dataModelUpdate", "deleteSurface"}
def extract_a2ui_json_objects(text: str) -> list[dict[str, Any]]:
"""Extract JSON objects containing A2UI keys from text.
Uses ``json.JSONDecoder.raw_decode`` for robust parsing that correctly
handles braces inside string literals.
"""
decoder = json.JSONDecoder()
results: list[dict[str, Any]] = []
idx = 0
while idx < len(text):
idx = text.find("{", idx)
if idx == -1:
break
try:
obj, end_idx = decoder.raw_decode(text, idx)
if isinstance(obj, dict) and _A2UI_KEYS & obj.keys():
results.append(obj)
idx = end_idx
except json.JSONDecodeError:
idx += 1
return results

View File

@@ -0,0 +1,76 @@
"""System prompt generation for A2UI-capable agents."""
from __future__ import annotations
import json
from crewai.a2a.extensions.a2ui.catalog import STANDARD_CATALOG_COMPONENTS
from crewai.a2a.extensions.a2ui.schema import load_schema
from crewai.a2a.extensions.a2ui.server_extension import A2UI_EXTENSION_URI
def build_a2ui_system_prompt(
catalog_id: str | None = None,
allowed_components: list[str] | None = None,
) -> str:
"""Build a system prompt fragment instructing the LLM to produce A2UI output.
The prompt describes the A2UI message format, available components, and
data binding rules. It includes the resolved schema so the LLM can
generate structured output.
Args:
catalog_id: Catalog identifier to reference. Defaults to the
standard catalog version derived from ``A2UI_EXTENSION_URI``.
allowed_components: Subset of component names to expose. When
``None``, all standard catalog components are available.
Returns:
A system prompt string to append to the agent's instructions.
"""
components = sorted(
allowed_components
if allowed_components is not None
else STANDARD_CATALOG_COMPONENTS
)
catalog_label = catalog_id or f"standard ({A2UI_EXTENSION_URI.rsplit('/', 1)[-1]})"
resolved_schema = load_schema("server_to_client_with_standard_catalog")
schema_json = json.dumps(resolved_schema, indent=2)
return f"""\
<A2UI_INSTRUCTIONS>
You can generate rich, declarative UI by emitting A2UI JSON messages.
CATALOG: {catalog_label}
AVAILABLE COMPONENTS: {", ".join(components)}
MESSAGE TYPES (emit exactly ONE per message):
- beginRendering: Initialize a new surface with a root component and optional styles.
- surfaceUpdate: Send/update components for a surface. Each component has a unique id \
and a "component" wrapper containing exactly one component-type key.
- dataModelUpdate: Update the data model for a surface. Data entries have a key and \
one typed value (valueString, valueNumber, valueBoolean, valueMap).
- deleteSurface: Remove a surface.
DATA BINDING:
- Use {{"literalString": "..."}} for inline string values.
- Use {{"literalNumber": ...}} for inline numeric values.
- Use {{"literalBoolean": ...}} for inline boolean values.
- Use {{"literalArray": ["...", "..."]}} for inline array values.
- Use {{"path": "/data/model/path"}} to bind to data model values.
ACTIONS:
- Interactive components (Button, etc.) have an "action" with a "name" and optional \
"context" array of key/value pairs.
- Values in action context can use data binding (path or literal).
OUTPUT FORMAT:
Emit each A2UI message as a valid JSON object. When generating UI, produce a \
beginRendering message first, then surfaceUpdate messages with components, and \
optionally dataModelUpdate messages to populate data-bound values.
SCHEMA:
{schema_json}
</A2UI_INSTRUCTIONS>"""

View File

@@ -0,0 +1,48 @@
"""Schema loading utilities for vendored A2UI JSON schemas."""
from __future__ import annotations
import json
from pathlib import Path
from typing import Any
_SCHEMA_DIR = Path(__file__).parent / "v0_8"
_SCHEMA_CACHE: dict[str, dict[str, Any]] = {}
SCHEMA_NAMES: frozenset[str] = frozenset(
{
"server_to_client",
"client_to_server",
"standard_catalog_definition",
"server_to_client_with_standard_catalog",
}
)
def load_schema(name: str) -> dict[str, Any]:
"""Load a vendored A2UI JSON schema by name.
Args:
name: Schema name without extension (e.g. ``"server_to_client"``).
Returns:
Parsed JSON schema dict.
Raises:
ValueError: If the schema name is not recognized.
FileNotFoundError: If the schema file is missing from the package.
"""
if name not in SCHEMA_NAMES:
raise ValueError(f"Unknown schema {name!r}. Available: {sorted(SCHEMA_NAMES)}")
if name in _SCHEMA_CACHE:
return _SCHEMA_CACHE[name]
path = _SCHEMA_DIR / f"{name}.json"
with path.open() as f:
schema: dict[str, Any] = json.load(f)
_SCHEMA_CACHE[name] = schema
return schema

View File

@@ -0,0 +1,53 @@
{
"title": "A2UI (Agent to UI) Client-to-Server Event Schema",
"description": "Describes a JSON payload for a client-to-server event message.",
"type": "object",
"minProperties": 1,
"maxProperties": 1,
"properties": {
"userAction": {
"type": "object",
"description": "Reports a user-initiated action from a component.",
"properties": {
"name": {
"type": "string",
"description": "The name of the action, taken from the component's action.name property."
},
"surfaceId": {
"type": "string",
"description": "The id of the surface where the event originated."
},
"sourceComponentId": {
"type": "string",
"description": "The id of the component that triggered the event."
},
"timestamp": {
"type": "string",
"format": "date-time",
"description": "An ISO 8601 timestamp of when the event occurred."
},
"context": {
"type": "object",
"description": "A JSON object containing the key-value pairs from the component's action.context, after resolving all data bindings.",
"additionalProperties": true
}
},
"required": [
"name",
"surfaceId",
"sourceComponentId",
"timestamp",
"context"
]
},
"error": {
"type": "object",
"description": "Reports a client-side error. The content is flexible.",
"additionalProperties": true
}
},
"oneOf": [
{ "required": ["userAction"] },
{ "required": ["error"] }
]
}

View File

@@ -0,0 +1,148 @@
{
"title": "A2UI Message Schema",
"description": "Describes a JSON payload for an A2UI (Agent to UI) message, which is used to dynamically construct and update user interfaces. A message MUST contain exactly ONE of the action properties: 'beginRendering', 'surfaceUpdate', 'dataModelUpdate', or 'deleteSurface'.",
"type": "object",
"additionalProperties": false,
"properties": {
"beginRendering": {
"type": "object",
"description": "Signals the client to begin rendering a surface with a root component and specific styles.",
"additionalProperties": false,
"properties": {
"surfaceId": {
"type": "string",
"description": "The unique identifier for the UI surface to be rendered."
},
"catalogId": {
"type": "string",
"description": "The identifier of the component catalog to use for this surface. If omitted, the client MUST default to the standard catalog for this A2UI version (https://a2ui.org/specification/v0_8/standard_catalog_definition.json)."
},
"root": {
"type": "string",
"description": "The ID of the root component to render."
},
"styles": {
"type": "object",
"description": "Styling information for the UI.",
"additionalProperties": true
}
},
"required": ["root", "surfaceId"]
},
"surfaceUpdate": {
"type": "object",
"description": "Updates a surface with a new set of components.",
"additionalProperties": false,
"properties": {
"surfaceId": {
"type": "string",
"description": "The unique identifier for the UI surface to be updated. If you are adding a new surface this *must* be a new, unique identified that has never been used for any existing surfaces shown."
},
"components": {
"type": "array",
"description": "A list containing all UI components for the surface.",
"minItems": 1,
"items": {
"type": "object",
"description": "Represents a *single* component in a UI widget tree. This component could be one of many supported types.",
"additionalProperties": false,
"properties": {
"id": {
"type": "string",
"description": "The unique identifier for this component."
},
"weight": {
"type": "number",
"description": "The relative weight of this component within a Row or Column. This corresponds to the CSS 'flex-grow' property. Note: this may ONLY be set when the component is a direct descendant of a Row or Column."
},
"component": {
"type": "object",
"description": "A wrapper object that MUST contain exactly one key, which is the name of the component type. The value is an object containing the properties for that specific component.",
"additionalProperties": true
}
},
"required": ["id", "component"]
}
}
},
"required": ["surfaceId", "components"]
},
"dataModelUpdate": {
"type": "object",
"description": "Updates the data model for a surface.",
"additionalProperties": false,
"properties": {
"surfaceId": {
"type": "string",
"description": "The unique identifier for the UI surface this data model update applies to."
},
"path": {
"type": "string",
"description": "An optional path to a location within the data model (e.g., '/user/name'). If omitted, or set to '/', the entire data model will be replaced."
},
"contents": {
"type": "array",
"description": "An array of data entries. Each entry must contain a 'key' and exactly one corresponding typed 'value*' property.",
"items": {
"type": "object",
"description": "A single data entry. Exactly one 'value*' property should be provided alongside the key.",
"additionalProperties": false,
"properties": {
"key": {
"type": "string",
"description": "The key for this data entry."
},
"valueString": {
"type": "string"
},
"valueNumber": {
"type": "number"
},
"valueBoolean": {
"type": "boolean"
},
"valueMap": {
"description": "Represents a map as an adjacency list.",
"type": "array",
"items": {
"type": "object",
"description": "One entry in the map. Exactly one 'value*' property should be provided alongside the key.",
"additionalProperties": false,
"properties": {
"key": {
"type": "string"
},
"valueString": {
"type": "string"
},
"valueNumber": {
"type": "number"
},
"valueBoolean": {
"type": "boolean"
}
},
"required": ["key"]
}
}
},
"required": ["key"]
}
}
},
"required": ["contents", "surfaceId"]
},
"deleteSurface": {
"type": "object",
"description": "Signals the client to delete the surface identified by 'surfaceId'.",
"additionalProperties": false,
"properties": {
"surfaceId": {
"type": "string",
"description": "The unique identifier for the UI surface to be deleted."
}
},
"required": ["surfaceId"]
}
}
}

View File

@@ -0,0 +1,823 @@
{
"title": "A2UI Message Schema",
"description": "Describes a JSON payload for an A2UI (Agent to UI) message, which is used to dynamically construct and update user interfaces. A message MUST contain exactly ONE of the action properties: 'beginRendering', 'surfaceUpdate', 'dataModelUpdate', or 'deleteSurface'.",
"type": "object",
"additionalProperties": false,
"properties": {
"beginRendering": {
"type": "object",
"description": "Signals the client to begin rendering a surface with a root component and specific styles.",
"additionalProperties": false,
"properties": {
"surfaceId": {
"type": "string",
"description": "The unique identifier for the UI surface to be rendered."
},
"root": {
"type": "string",
"description": "The ID of the root component to render."
},
"styles": {
"type": "object",
"description": "Styling information for the UI.",
"additionalProperties": false,
"properties": {
"font": {
"type": "string",
"description": "The primary font for the UI."
},
"primaryColor": {
"type": "string",
"description": "The primary UI color as a hexadecimal code (e.g., '#00BFFF').",
"pattern": "^#[0-9a-fA-F]{6}$"
}
}
}
},
"required": ["root", "surfaceId"]
},
"surfaceUpdate": {
"type": "object",
"description": "Updates a surface with a new set of components.",
"additionalProperties": false,
"properties": {
"surfaceId": {
"type": "string",
"description": "The unique identifier for the UI surface to be updated. If you are adding a new surface this *must* be a new, unique identified that has never been used for any existing surfaces shown."
},
"components": {
"type": "array",
"description": "A list containing all UI components for the surface.",
"minItems": 1,
"items": {
"type": "object",
"description": "Represents a *single* component in a UI widget tree. This component could be one of many supported types.",
"additionalProperties": false,
"properties": {
"id": {
"type": "string",
"description": "The unique identifier for this component."
},
"weight": {
"type": "number",
"description": "The relative weight of this component within a Row or Column. This corresponds to the CSS 'flex-grow' property. Note: this may ONLY be set when the component is a direct descendant of a Row or Column."
},
"component": {
"type": "object",
"description": "A wrapper object that MUST contain exactly one key, which is the name of the component type (e.g., 'Heading'). The value is an object containing the properties for that specific component.",
"additionalProperties": false,
"properties": {
"Text": {
"type": "object",
"additionalProperties": false,
"properties": {
"text": {
"type": "object",
"description": "The text content to display. This can be a literal string or a reference to a value in the data model ('path', e.g., '/doc/title'). While simple Markdown formatting is supported (i.e. without HTML, images, or links), utilizing dedicated UI components is generally preferred for a richer and more structured presentation.",
"additionalProperties": false,
"properties": {
"literalString": {
"type": "string"
},
"path": {
"type": "string"
}
}
},
"usageHint": {
"type": "string",
"description": "A hint for the base text style. One of:\n- `h1`: Largest heading.\n- `h2`: Second largest heading.\n- `h3`: Third largest heading.\n- `h4`: Fourth largest heading.\n- `h5`: Fifth largest heading.\n- `caption`: Small text for captions.\n- `body`: Standard body text.",
"enum": [
"h1",
"h2",
"h3",
"h4",
"h5",
"caption",
"body"
]
}
},
"required": ["text"]
},
"Image": {
"type": "object",
"additionalProperties": false,
"properties": {
"url": {
"type": "object",
"description": "The URL of the image to display. This can be a literal string ('literal') or a reference to a value in the data model ('path', e.g. '/thumbnail/url').",
"additionalProperties": false,
"properties": {
"literalString": {
"type": "string"
},
"path": {
"type": "string"
}
}
},
"fit": {
"type": "string",
"description": "Specifies how the image should be resized to fit its container. This corresponds to the CSS 'object-fit' property.",
"enum": [
"contain",
"cover",
"fill",
"none",
"scale-down"
]
},
"usageHint": {
"type": "string",
"description": "A hint for the image size and style. One of:\n- `icon`: Small square icon.\n- `avatar`: Circular avatar image.\n- `smallFeature`: Small feature image.\n- `mediumFeature`: Medium feature image.\n- `largeFeature`: Large feature image.\n- `header`: Full-width, full bleed, header image.",
"enum": [
"icon",
"avatar",
"smallFeature",
"mediumFeature",
"largeFeature",
"header"
]
}
},
"required": ["url"]
},
"Icon": {
"type": "object",
"additionalProperties": false,
"properties": {
"name": {
"type": "object",
"description": "The name of the icon to display. This can be a literal string or a reference to a value in the data model ('path', e.g. '/form/submit').",
"additionalProperties": false,
"properties": {
"literalString": {
"type": "string",
"enum": [
"accountCircle",
"add",
"arrowBack",
"arrowForward",
"attachFile",
"calendarToday",
"call",
"camera",
"check",
"close",
"delete",
"download",
"edit",
"event",
"error",
"favorite",
"favoriteOff",
"folder",
"help",
"home",
"info",
"locationOn",
"lock",
"lockOpen",
"mail",
"menu",
"moreVert",
"moreHoriz",
"notificationsOff",
"notifications",
"payment",
"person",
"phone",
"photo",
"print",
"refresh",
"search",
"send",
"settings",
"share",
"shoppingCart",
"star",
"starHalf",
"starOff",
"upload",
"visibility",
"visibilityOff",
"warning"
]
},
"path": {
"type": "string"
}
}
}
},
"required": ["name"]
},
"Video": {
"type": "object",
"additionalProperties": false,
"properties": {
"url": {
"type": "object",
"description": "The URL of the video to display. This can be a literal string or a reference to a value in the data model ('path', e.g. '/video/url').",
"additionalProperties": false,
"properties": {
"literalString": {
"type": "string"
},
"path": {
"type": "string"
}
}
}
},
"required": ["url"]
},
"AudioPlayer": {
"type": "object",
"additionalProperties": false,
"properties": {
"url": {
"type": "object",
"description": "The URL of the audio to be played. This can be a literal string ('literal') or a reference to a value in the data model ('path', e.g. '/song/url').",
"additionalProperties": false,
"properties": {
"literalString": {
"type": "string"
},
"path": {
"type": "string"
}
}
},
"description": {
"type": "object",
"description": "A description of the audio, such as a title or summary. This can be a literal string or a reference to a value in the data model ('path', e.g. '/song/title').",
"additionalProperties": false,
"properties": {
"literalString": {
"type": "string"
},
"path": {
"type": "string"
}
}
}
},
"required": ["url"]
},
"Row": {
"type": "object",
"additionalProperties": false,
"properties": {
"children": {
"type": "object",
"description": "Defines the children. Use 'explicitList' for a fixed set of children, or 'template' to generate children from a data list.",
"additionalProperties": false,
"properties": {
"explicitList": {
"type": "array",
"items": {
"type": "string"
}
},
"template": {
"type": "object",
"description": "A template for generating a dynamic list of children from a data model list. `componentId` is the component to use as a template, and `dataBinding` is the path to the map of components in the data model. Values in the map will define the list of children.",
"additionalProperties": false,
"properties": {
"componentId": {
"type": "string"
},
"dataBinding": {
"type": "string"
}
},
"required": ["componentId", "dataBinding"]
}
}
},
"distribution": {
"type": "string",
"description": "Defines the arrangement of children along the main axis (horizontally). This corresponds to the CSS 'justify-content' property.",
"enum": [
"center",
"end",
"spaceAround",
"spaceBetween",
"spaceEvenly",
"start"
]
},
"alignment": {
"type": "string",
"description": "Defines the alignment of children along the cross axis (vertically). This corresponds to the CSS 'align-items' property.",
"enum": ["start", "center", "end", "stretch"]
}
},
"required": ["children"]
},
"Column": {
"type": "object",
"additionalProperties": false,
"properties": {
"children": {
"type": "object",
"description": "Defines the children. Use 'explicitList' for a fixed set of children, or 'template' to generate children from a data list.",
"additionalProperties": false,
"properties": {
"explicitList": {
"type": "array",
"items": {
"type": "string"
}
},
"template": {
"type": "object",
"description": "A template for generating a dynamic list of children from a data model list. `componentId` is the component to use as a template, and `dataBinding` is the path to the map of components in the data model. Values in the map will define the list of children.",
"additionalProperties": false,
"properties": {
"componentId": {
"type": "string"
},
"dataBinding": {
"type": "string"
}
},
"required": ["componentId", "dataBinding"]
}
}
},
"distribution": {
"type": "string",
"description": "Defines the arrangement of children along the main axis (vertically). This corresponds to the CSS 'justify-content' property.",
"enum": [
"start",
"center",
"end",
"spaceBetween",
"spaceAround",
"spaceEvenly"
]
},
"alignment": {
"type": "string",
"description": "Defines the alignment of children along the cross axis (horizontally). This corresponds to the CSS 'align-items' property.",
"enum": ["center", "end", "start", "stretch"]
}
},
"required": ["children"]
},
"List": {
"type": "object",
"additionalProperties": false,
"properties": {
"children": {
"type": "object",
"description": "Defines the children. Use 'explicitList' for a fixed set of children, or 'template' to generate children from a data list.",
"additionalProperties": false,
"properties": {
"explicitList": {
"type": "array",
"items": {
"type": "string"
}
},
"template": {
"type": "object",
"description": "A template for generating a dynamic list of children from a data model list. `componentId` is the component to use as a template, and `dataBinding` is the path to the map of components in the data model. Values in the map will define the list of children.",
"additionalProperties": false,
"properties": {
"componentId": {
"type": "string"
},
"dataBinding": {
"type": "string"
}
},
"required": ["componentId", "dataBinding"]
}
}
},
"direction": {
"type": "string",
"description": "The direction in which the list items are laid out.",
"enum": ["vertical", "horizontal"]
},
"alignment": {
"type": "string",
"description": "Defines the alignment of children along the cross axis.",
"enum": ["start", "center", "end", "stretch"]
}
},
"required": ["children"]
},
"Card": {
"type": "object",
"additionalProperties": false,
"properties": {
"child": {
"type": "string",
"description": "The ID of the component to be rendered inside the card."
}
},
"required": ["child"]
},
"Tabs": {
"type": "object",
"additionalProperties": false,
"properties": {
"tabItems": {
"type": "array",
"description": "An array of objects, where each object defines a tab with a title and a child component.",
"items": {
"type": "object",
"additionalProperties": false,
"properties": {
"title": {
"type": "object",
"description": "The tab title. Defines the value as either a literal value or a path to data model value (e.g. '/options/title').",
"additionalProperties": false,
"properties": {
"literalString": {
"type": "string"
},
"path": {
"type": "string"
}
}
},
"child": {
"type": "string"
}
},
"required": ["title", "child"]
}
}
},
"required": ["tabItems"]
},
"Divider": {
"type": "object",
"additionalProperties": false,
"properties": {
"axis": {
"type": "string",
"description": "The orientation of the divider.",
"enum": ["horizontal", "vertical"]
}
}
},
"Modal": {
"type": "object",
"additionalProperties": false,
"properties": {
"entryPointChild": {
"type": "string",
"description": "The ID of the component that opens the modal when interacted with (e.g., a button)."
},
"contentChild": {
"type": "string",
"description": "The ID of the component to be displayed inside the modal."
}
},
"required": ["entryPointChild", "contentChild"]
},
"Button": {
"type": "object",
"additionalProperties": false,
"properties": {
"child": {
"type": "string",
"description": "The ID of the component to display in the button, typically a Text component."
},
"primary": {
"type": "boolean",
"description": "Indicates if this button should be styled as the primary action."
},
"action": {
"type": "object",
"description": "The client-side action to be dispatched when the button is clicked. It includes the action's name and an optional context payload.",
"additionalProperties": false,
"properties": {
"name": {
"type": "string"
},
"context": {
"type": "array",
"items": {
"type": "object",
"additionalProperties": false,
"properties": {
"key": {
"type": "string"
},
"value": {
"type": "object",
"description": "Defines the value to be included in the context as either a literal value or a path to a data model value (e.g. '/user/name').",
"additionalProperties": false,
"properties": {
"path": {
"type": "string"
},
"literalString": {
"type": "string"
},
"literalNumber": {
"type": "number"
},
"literalBoolean": {
"type": "boolean"
}
}
}
},
"required": ["key", "value"]
}
}
},
"required": ["name"]
}
},
"required": ["child", "action"]
},
"CheckBox": {
"type": "object",
"additionalProperties": false,
"properties": {
"label": {
"type": "object",
"description": "The text to display next to the checkbox. Defines the value as either a literal value or a path to data model ('path', e.g. '/option/label').",
"additionalProperties": false,
"properties": {
"literalString": {
"type": "string"
},
"path": {
"type": "string"
}
}
},
"value": {
"type": "object",
"description": "The current state of the checkbox (true for checked, false for unchecked). This can be a literal boolean ('literalBoolean') or a reference to a value in the data model ('path', e.g. '/filter/open').",
"additionalProperties": false,
"properties": {
"literalBoolean": {
"type": "boolean"
},
"path": {
"type": "string"
}
}
}
},
"required": ["label", "value"]
},
"TextField": {
"type": "object",
"additionalProperties": false,
"properties": {
"label": {
"type": "object",
"description": "The text label for the input field. This can be a literal string or a reference to a value in the data model ('path, e.g. '/user/name').",
"additionalProperties": false,
"properties": {
"literalString": {
"type": "string"
},
"path": {
"type": "string"
}
}
},
"text": {
"type": "object",
"description": "The value of the text field. This can be a literal string or a reference to a value in the data model ('path', e.g. '/user/name').",
"additionalProperties": false,
"properties": {
"literalString": {
"type": "string"
},
"path": {
"type": "string"
}
}
},
"textFieldType": {
"type": "string",
"description": "The type of input field to display.",
"enum": [
"date",
"longText",
"number",
"shortText",
"obscured"
]
},
"validationRegexp": {
"type": "string",
"description": "A regular expression used for client-side validation of the input."
}
},
"required": ["label"]
},
"DateTimeInput": {
"type": "object",
"additionalProperties": false,
"properties": {
"value": {
"type": "object",
"description": "The selected date and/or time value in ISO 8601 format. This can be a literal string ('literalString') or a reference to a value in the data model ('path', e.g. '/user/dob').",
"additionalProperties": false,
"properties": {
"literalString": {
"type": "string"
},
"path": {
"type": "string"
}
}
},
"enableDate": {
"type": "boolean",
"description": "If true, allows the user to select a date."
},
"enableTime": {
"type": "boolean",
"description": "If true, allows the user to select a time."
}
},
"required": ["value"]
},
"MultipleChoice": {
"type": "object",
"additionalProperties": false,
"properties": {
"selections": {
"type": "object",
"description": "The currently selected values for the component. This can be a literal array of strings or a path to an array in the data model('path', e.g. '/hotel/options').",
"additionalProperties": false,
"properties": {
"literalArray": {
"type": "array",
"items": {
"type": "string"
}
},
"path": {
"type": "string"
}
}
},
"options": {
"type": "array",
"description": "An array of available options for the user to choose from.",
"items": {
"type": "object",
"additionalProperties": false,
"properties": {
"label": {
"type": "object",
"description": "The text to display for this option. This can be a literal string or a reference to a value in the data model (e.g. '/option/label').",
"additionalProperties": false,
"properties": {
"literalString": {
"type": "string"
},
"path": {
"type": "string"
}
}
},
"value": {
"type": "string",
"description": "The value to be associated with this option when selected."
}
},
"required": ["label", "value"]
}
},
"maxAllowedSelections": {
"type": "integer",
"description": "The maximum number of options that the user is allowed to select."
}
},
"required": ["selections", "options"]
},
"Slider": {
"type": "object",
"additionalProperties": false,
"properties": {
"value": {
"type": "object",
"description": "The current value of the slider. This can be a literal number ('literalNumber') or a reference to a value in the data model ('path', e.g. '/restaurant/cost').",
"additionalProperties": false,
"properties": {
"literalNumber": {
"type": "number"
},
"path": {
"type": "string"
}
}
},
"minValue": {
"type": "number",
"description": "The minimum value of the slider."
},
"maxValue": {
"type": "number",
"description": "The maximum value of the slider."
}
},
"required": ["value"]
}
}
}
},
"required": ["id", "component"]
}
}
},
"required": ["surfaceId", "components"]
},
"dataModelUpdate": {
"type": "object",
"description": "Updates the data model for a surface.",
"additionalProperties": false,
"properties": {
"surfaceId": {
"type": "string",
"description": "The unique identifier for the UI surface this data model update applies to."
},
"path": {
"type": "string",
"description": "An optional path to a location within the data model (e.g., '/user/name'). If omitted, or set to '/', the entire data model will be replaced."
},
"contents": {
"type": "array",
"description": "An array of data entries. Each entry must contain a 'key' and exactly one corresponding typed 'value*' property.",
"items": {
"type": "object",
"description": "A single data entry. Exactly one 'value*' property should be provided alongside the key.",
"additionalProperties": false,
"properties": {
"key": {
"type": "string",
"description": "The key for this data entry."
},
"valueString": {
"type": "string"
},
"valueNumber": {
"type": "number"
},
"valueBoolean": {
"type": "boolean"
},
"valueMap": {
"description": "Represents a map as an adjacency list.",
"type": "array",
"items": {
"type": "object",
"description": "One entry in the map. Exactly one 'value*' property should be provided alongside the key.",
"additionalProperties": false,
"properties": {
"key": {
"type": "string"
},
"valueString": {
"type": "string"
},
"valueNumber": {
"type": "number"
},
"valueBoolean": {
"type": "boolean"
}
},
"required": ["key"]
}
}
},
"required": ["key"]
}
}
},
"required": ["contents", "surfaceId"]
},
"deleteSurface": {
"type": "object",
"description": "Signals the client to delete the surface identified by 'surfaceId'.",
"additionalProperties": false,
"properties": {
"surfaceId": {
"type": "string",
"description": "The unique identifier for the UI surface to be deleted."
}
},
"required": ["surfaceId"]
}
}
}

View File

@@ -0,0 +1,459 @@
{
"components": {
"Text": {
"type": "object",
"additionalProperties": false,
"properties": {
"text": {
"type": "object",
"description": "The text content to display. This can be a literal string or a reference to a value in the data model ('path', e.g., '/doc/title'). While simple Markdown formatting is supported (i.e. without HTML, images, or links), utilizing dedicated UI components is generally preferred for a richer and more structured presentation.",
"additionalProperties": false,
"properties": {
"literalString": { "type": "string" },
"path": { "type": "string" }
}
},
"usageHint": {
"type": "string",
"description": "A hint for the base text style.",
"enum": ["h1", "h2", "h3", "h4", "h5", "caption", "body"]
}
},
"required": ["text"]
},
"Image": {
"type": "object",
"additionalProperties": false,
"properties": {
"url": {
"type": "object",
"description": "The URL of the image to display.",
"additionalProperties": false,
"properties": {
"literalString": { "type": "string" },
"path": { "type": "string" }
}
},
"fit": {
"type": "string",
"description": "Specifies how the image should be resized to fit its container.",
"enum": ["contain", "cover", "fill", "none", "scale-down"]
},
"usageHint": {
"type": "string",
"description": "A hint for the image size and style.",
"enum": ["icon", "avatar", "smallFeature", "mediumFeature", "largeFeature", "header"]
}
},
"required": ["url"]
},
"Icon": {
"type": "object",
"additionalProperties": false,
"properties": {
"name": {
"type": "object",
"description": "The name of the icon to display.",
"additionalProperties": false,
"properties": {
"literalString": {
"type": "string",
"enum": [
"accountCircle", "add", "arrowBack", "arrowForward", "attachFile",
"calendarToday", "call", "camera", "check", "close", "delete",
"download", "edit", "event", "error", "favorite", "favoriteOff",
"folder", "help", "home", "info", "locationOn", "lock", "lockOpen",
"mail", "menu", "moreVert", "moreHoriz", "notificationsOff",
"notifications", "payment", "person", "phone", "photo", "print",
"refresh", "search", "send", "settings", "share", "shoppingCart",
"star", "starHalf", "starOff", "upload", "visibility",
"visibilityOff", "warning"
]
},
"path": { "type": "string" }
}
}
},
"required": ["name"]
},
"Video": {
"type": "object",
"additionalProperties": false,
"properties": {
"url": {
"type": "object",
"description": "The URL of the video to display.",
"additionalProperties": false,
"properties": {
"literalString": { "type": "string" },
"path": { "type": "string" }
}
}
},
"required": ["url"]
},
"AudioPlayer": {
"type": "object",
"additionalProperties": false,
"properties": {
"url": {
"type": "object",
"description": "The URL of the audio to be played.",
"additionalProperties": false,
"properties": {
"literalString": { "type": "string" },
"path": { "type": "string" }
}
},
"description": {
"type": "object",
"description": "A description of the audio, such as a title or summary.",
"additionalProperties": false,
"properties": {
"literalString": { "type": "string" },
"path": { "type": "string" }
}
}
},
"required": ["url"]
},
"Row": {
"type": "object",
"additionalProperties": false,
"properties": {
"children": {
"type": "object",
"description": "Defines the children. Use 'explicitList' for a fixed set of children, or 'template' to generate children from a data list.",
"additionalProperties": false,
"properties": {
"explicitList": { "type": "array", "items": { "type": "string" } },
"template": {
"type": "object",
"additionalProperties": false,
"properties": {
"componentId": { "type": "string" },
"dataBinding": { "type": "string" }
},
"required": ["componentId", "dataBinding"]
}
}
},
"distribution": {
"type": "string",
"enum": ["center", "end", "spaceAround", "spaceBetween", "spaceEvenly", "start"]
},
"alignment": {
"type": "string",
"enum": ["start", "center", "end", "stretch"]
}
},
"required": ["children"]
},
"Column": {
"type": "object",
"additionalProperties": false,
"properties": {
"children": {
"type": "object",
"description": "Defines the children. Use 'explicitList' for a fixed set of children, or 'template' to generate children from a data list.",
"additionalProperties": false,
"properties": {
"explicitList": { "type": "array", "items": { "type": "string" } },
"template": {
"type": "object",
"additionalProperties": false,
"properties": {
"componentId": { "type": "string" },
"dataBinding": { "type": "string" }
},
"required": ["componentId", "dataBinding"]
}
}
},
"distribution": {
"type": "string",
"enum": ["start", "center", "end", "spaceBetween", "spaceAround", "spaceEvenly"]
},
"alignment": {
"type": "string",
"enum": ["center", "end", "start", "stretch"]
}
},
"required": ["children"]
},
"List": {
"type": "object",
"additionalProperties": false,
"properties": {
"children": {
"type": "object",
"description": "Defines the children. Use 'explicitList' for a fixed set of children, or 'template' to generate children from a data list.",
"additionalProperties": false,
"properties": {
"explicitList": { "type": "array", "items": { "type": "string" } },
"template": {
"type": "object",
"additionalProperties": false,
"properties": {
"componentId": { "type": "string" },
"dataBinding": { "type": "string" }
},
"required": ["componentId", "dataBinding"]
}
}
},
"direction": {
"type": "string",
"enum": ["vertical", "horizontal"]
},
"alignment": {
"type": "string",
"enum": ["start", "center", "end", "stretch"]
}
},
"required": ["children"]
},
"Card": {
"type": "object",
"additionalProperties": false,
"properties": {
"child": {
"type": "string",
"description": "The ID of the component to be rendered inside the card."
}
},
"required": ["child"]
},
"Tabs": {
"type": "object",
"additionalProperties": false,
"properties": {
"tabItems": {
"type": "array",
"items": {
"type": "object",
"additionalProperties": false,
"properties": {
"title": {
"type": "object",
"additionalProperties": false,
"properties": {
"literalString": { "type": "string" },
"path": { "type": "string" }
}
},
"child": { "type": "string" }
},
"required": ["title", "child"]
}
}
},
"required": ["tabItems"]
},
"Divider": {
"type": "object",
"additionalProperties": false,
"properties": {
"axis": {
"type": "string",
"enum": ["horizontal", "vertical"]
}
}
},
"Modal": {
"type": "object",
"additionalProperties": false,
"properties": {
"entryPointChild": {
"type": "string",
"description": "The ID of the component that opens the modal when interacted with."
},
"contentChild": {
"type": "string",
"description": "The ID of the component to be displayed inside the modal."
}
},
"required": ["entryPointChild", "contentChild"]
},
"Button": {
"type": "object",
"additionalProperties": false,
"properties": {
"child": {
"type": "string",
"description": "The ID of the component to display in the button."
},
"primary": {
"type": "boolean",
"description": "Indicates if this button should be styled as the primary action."
},
"action": {
"type": "object",
"additionalProperties": false,
"properties": {
"name": { "type": "string" },
"context": {
"type": "array",
"items": {
"type": "object",
"additionalProperties": false,
"properties": {
"key": { "type": "string" },
"value": {
"type": "object",
"additionalProperties": false,
"properties": {
"path": { "type": "string" },
"literalString": { "type": "string" },
"literalNumber": { "type": "number" },
"literalBoolean": { "type": "boolean" }
}
}
},
"required": ["key", "value"]
}
}
},
"required": ["name"]
}
},
"required": ["child", "action"]
},
"CheckBox": {
"type": "object",
"additionalProperties": false,
"properties": {
"label": {
"type": "object",
"additionalProperties": false,
"properties": {
"literalString": { "type": "string" },
"path": { "type": "string" }
}
},
"value": {
"type": "object",
"additionalProperties": false,
"properties": {
"literalBoolean": { "type": "boolean" },
"path": { "type": "string" }
}
}
},
"required": ["label", "value"]
},
"TextField": {
"type": "object",
"additionalProperties": false,
"properties": {
"label": {
"type": "object",
"additionalProperties": false,
"properties": {
"literalString": { "type": "string" },
"path": { "type": "string" }
}
},
"text": {
"type": "object",
"additionalProperties": false,
"properties": {
"literalString": { "type": "string" },
"path": { "type": "string" }
}
},
"textFieldType": {
"type": "string",
"enum": ["date", "longText", "number", "shortText", "obscured"]
},
"validationRegexp": { "type": "string" }
},
"required": ["label"]
},
"DateTimeInput": {
"type": "object",
"additionalProperties": false,
"properties": {
"value": {
"type": "object",
"additionalProperties": false,
"properties": {
"literalString": { "type": "string" },
"path": { "type": "string" }
}
},
"enableDate": { "type": "boolean" },
"enableTime": { "type": "boolean" }
},
"required": ["value"]
},
"MultipleChoice": {
"type": "object",
"additionalProperties": false,
"properties": {
"selections": {
"type": "object",
"additionalProperties": false,
"properties": {
"literalArray": { "type": "array", "items": { "type": "string" } },
"path": { "type": "string" }
}
},
"options": {
"type": "array",
"items": {
"type": "object",
"additionalProperties": false,
"properties": {
"label": {
"type": "object",
"additionalProperties": false,
"properties": {
"literalString": { "type": "string" },
"path": { "type": "string" }
}
},
"value": { "type": "string" }
},
"required": ["label", "value"]
}
},
"maxAllowedSelections": { "type": "integer" },
"variant": {
"type": "string",
"enum": ["checkbox", "chips"]
},
"filterable": { "type": "boolean" }
},
"required": ["selections", "options"]
},
"Slider": {
"type": "object",
"additionalProperties": false,
"properties": {
"value": {
"type": "object",
"additionalProperties": false,
"properties": {
"literalNumber": { "type": "number" },
"path": { "type": "string" }
}
},
"minValue": { "type": "number" },
"maxValue": { "type": "number" }
},
"required": ["value"]
}
},
"styles": {
"font": {
"type": "string",
"description": "The primary font for the UI."
},
"primaryColor": {
"type": "string",
"description": "The primary UI color as a hexadecimal code (e.g., '#00BFFF').",
"pattern": "^#[0-9a-fA-F]{6}$"
}
}
}

View File

@@ -0,0 +1,125 @@
"""A2UI server extension for the A2A protocol."""
from __future__ import annotations
import logging
from typing import Any
from crewai.a2a.extensions.a2ui.models import A2UIResponse, extract_a2ui_json_objects
from crewai.a2a.extensions.a2ui.validator import (
A2UIValidationError,
validate_a2ui_message,
)
from crewai.a2a.extensions.server import ExtensionContext, ServerExtension
logger = logging.getLogger(__name__)
A2UI_MIME_TYPE = "application/json+a2ui"
A2UI_EXTENSION_URI = "https://a2ui.org/a2a-extension/a2ui/v0.8"
class A2UIServerExtension(ServerExtension):
"""A2A server extension that enables A2UI declarative UI generation.
When activated by a client that declares A2UI v0.8 support,
this extension:
* Negotiates catalog preferences during ``on_request``.
* Wraps A2UI messages in the agent response as A2A DataParts with
``application/json+a2ui`` MIME type during ``on_response``.
Example::
A2AServerConfig(
server_extensions=[A2UIServerExtension()],
default_output_modes=["text/plain", "application/json+a2ui"],
)
"""
uri: str = A2UI_EXTENSION_URI
required: bool = False
description: str = "A2UI declarative UI generation"
def __init__(
self,
catalog_ids: list[str] | None = None,
accept_inline_catalogs: bool = False,
) -> None:
"""Initialize the A2UI server extension.
Args:
catalog_ids: Catalog identifiers this server supports.
accept_inline_catalogs: Whether inline catalog definitions are accepted.
"""
self._catalog_ids = catalog_ids or []
self._accept_inline_catalogs = accept_inline_catalogs
@property
def params(self) -> dict[str, Any]:
"""Extension parameters advertised in the AgentCard."""
result: dict[str, Any] = {}
if self._catalog_ids:
result["supportedCatalogIds"] = self._catalog_ids
result["acceptsInlineCatalogs"] = self._accept_inline_catalogs
return result
async def on_request(self, context: ExtensionContext) -> None:
"""Extract A2UI catalog preferences from the client request.
Stores the negotiated catalog in ``context.state`` under
``"a2ui_catalog_id"`` for downstream use.
"""
if not self.is_active(context):
return
catalog_id = context.get_extension_metadata(self.uri, "catalogId")
if isinstance(catalog_id, str):
context.state["a2ui_catalog_id"] = catalog_id
elif self._catalog_ids:
context.state["a2ui_catalog_id"] = self._catalog_ids[0]
context.state["a2ui_active"] = True
async def on_response(self, context: ExtensionContext, result: Any) -> Any:
"""Wrap A2UI messages in the result as A2A DataParts.
Scans the result for A2UI JSON payloads and converts them into
DataParts with ``application/json+a2ui`` MIME type and A2UI metadata.
"""
if not context.state.get("a2ui_active"):
return result
if not isinstance(result, str):
return result
a2ui_messages = extract_a2ui_json_objects(result)
if not a2ui_messages:
return result
data_parts = [
part
for part in (_build_data_part(msg_data) for msg_data in a2ui_messages)
if part is not None
]
if not data_parts:
return result
return A2UIResponse(text=result, a2ui_parts=data_parts)
def _build_data_part(msg_data: dict[str, Any]) -> dict[str, Any] | None:
"""Validate a single A2UI message and wrap it as a DataPart dict."""
try:
validated = validate_a2ui_message(msg_data)
except A2UIValidationError:
logger.warning("Skipping invalid A2UI message in response", exc_info=True)
return None
return {
"kind": "data",
"data": validated.model_dump(by_alias=True, exclude_none=True),
"metadata": {
"mimeType": A2UI_MIME_TYPE,
},
}

View File

@@ -0,0 +1,59 @@
"""Validate A2UI message dicts via Pydantic models."""
from __future__ import annotations
from typing import Any
from pydantic import ValidationError
from crewai.a2a.extensions.a2ui.models import A2UIEvent, A2UIMessage
class A2UIValidationError(Exception):
"""Raised when an A2UI message fails validation."""
def __init__(self, message: str, errors: list[Any] | None = None) -> None:
super().__init__(message)
self.errors = errors or []
def validate_a2ui_message(data: dict[str, Any]) -> A2UIMessage:
"""Parse and validate an A2UI server-to-client message.
Args:
data: Raw message dict (JSON-decoded).
Returns:
Validated ``A2UIMessage`` instance.
Raises:
A2UIValidationError: If the data does not conform to the A2UI schema.
"""
try:
return A2UIMessage.model_validate(data)
except ValidationError as exc:
raise A2UIValidationError(
f"Invalid A2UI message: {exc.error_count()} validation error(s)",
errors=exc.errors(),
) from exc
def validate_a2ui_event(data: dict[str, Any]) -> A2UIEvent:
"""Parse and validate an A2UI client-to-server event.
Args:
data: Raw event dict (JSON-decoded).
Returns:
Validated ``A2UIEvent`` instance.
Raises:
A2UIValidationError: If the data does not conform to the A2UI event schema.
"""
try:
return A2UIEvent.model_validate(data)
except ValidationError as exc:
raise A2UIValidationError(
f"Invalid A2UI event: {exc.error_count()} validation error(s)",
errors=exc.errors(),
) from exc

View File

@@ -3,7 +3,7 @@
from __future__ import annotations
from collections.abc import AsyncIterator
from typing import TYPE_CHECKING, Any, TypedDict
from typing import TYPE_CHECKING, Any
import uuid
from a2a.client.errors import A2AClientHTTPError
@@ -18,7 +18,7 @@ from a2a.types import (
TaskStatusUpdateEvent,
TextPart,
)
from typing_extensions import NotRequired
from typing_extensions import NotRequired, TypedDict
from crewai.events.event_bus import crewai_event_bus
from crewai.events.types.a2a_events import (

View File

@@ -7,12 +7,11 @@ from typing import (
Any,
Literal,
Protocol,
TypedDict,
runtime_checkable,
)
from pydantic import BeforeValidator, HttpUrl, TypeAdapter
from typing_extensions import NotRequired
from typing_extensions import NotRequired, TypedDict
try:

View File

@@ -2,10 +2,11 @@
from __future__ import annotations
from typing import TYPE_CHECKING, Any, NamedTuple, Protocol, TypedDict
from typing import TYPE_CHECKING, Any, NamedTuple, Protocol
from pydantic import GetCoreSchemaHandler
from pydantic_core import CoreSchema, core_schema
from typing_extensions import TypedDict
class CommonParams(NamedTuple):

View File

@@ -28,6 +28,7 @@ APPLICATION_PDF: Literal["application/pdf"] = "application/pdf"
APPLICATION_OCTET_STREAM: Literal["application/octet-stream"] = (
"application/octet-stream"
)
APPLICATION_A2UI_JSON: Literal["application/json+a2ui"] = "application/json+a2ui"
DEFAULT_CLIENT_INPUT_MODES: Final[list[Literal["text/plain", "application/json"]]] = [
TEXT_PLAIN,
@@ -311,6 +312,10 @@ def get_part_content_type(part: Part) -> str:
if root.kind == "text":
return TEXT_PLAIN
if root.kind == "data":
metadata = root.metadata or {}
mime = metadata.get("mimeType", "")
if mime == APPLICATION_A2UI_JSON:
return APPLICATION_A2UI_JSON
return APPLICATION_JSON
if root.kind == "file":
return root.file.mime_type or APPLICATION_OCTET_STREAM

View File

@@ -10,7 +10,7 @@ from functools import wraps
import json
import logging
import os
from typing import TYPE_CHECKING, Any, ParamSpec, TypeVar, TypedDict, cast
from typing import TYPE_CHECKING, Any, ParamSpec, TypeVar, cast
from urllib.parse import urlparse
from a2a.server.agent_execution import RequestContext
@@ -38,6 +38,7 @@ from a2a.utils import (
from a2a.utils.errors import ServerError
from aiocache import SimpleMemoryCache, caches # type: ignore[import-untyped]
from pydantic import BaseModel
from typing_extensions import TypedDict
from crewai.a2a.utils.agent_card import _get_server_config
from crewai.a2a.utils.content_type import validate_message_parts

View File

@@ -0,0 +1,319 @@
"""Cross-validate A2UI Pydantic models against vendored JSON schemas.
Ensures the two validation sources stay in sync: representative payloads
must be accepted or rejected consistently by both the Pydantic models and
the JSON schemas.
"""
from __future__ import annotations
from typing import Any
import jsonschema
import pytest
from crewai.a2a.extensions.a2ui import catalog
from crewai.a2a.extensions.a2ui.models import A2UIEvent, A2UIMessage
from crewai.a2a.extensions.a2ui.schema import load_schema
SERVER_SCHEMA = load_schema("server_to_client")
CLIENT_SCHEMA = load_schema("client_to_server")
CATALOG_SCHEMA = load_schema("standard_catalog_definition")
def _json_schema_valid(schema: dict[str, Any], instance: dict[str, Any]) -> bool:
"""Return True if *instance* validates against *schema*."""
try:
jsonschema.validate(instance, schema)
return True
except jsonschema.ValidationError:
return False
def _pydantic_valid_message(data: dict[str, Any]) -> bool:
"""Return True if *data* validates as an A2UIMessage."""
try:
A2UIMessage.model_validate(data)
return True
except Exception:
return False
def _pydantic_valid_event(data: dict[str, Any]) -> bool:
"""Return True if *data* validates as an A2UIEvent."""
try:
A2UIEvent.model_validate(data)
return True
except Exception:
return False
# ---------------------------------------------------------------------------
# Valid server-to-client payloads
# ---------------------------------------------------------------------------
VALID_SERVER_MESSAGES: list[dict[str, Any]] = [
{
"beginRendering": {
"surfaceId": "s1",
"root": "root-col",
},
},
{
"beginRendering": {
"surfaceId": "s2",
"root": "root-col",
"catalogId": "standard (v0.8)",
"styles": {"primaryColor": "#FF0000", "font": "Roboto"},
},
},
{
"surfaceUpdate": {
"surfaceId": "s1",
"components": [
{
"id": "title",
"component": {
"Text": {"text": {"literalString": "Hello"}},
},
},
],
},
},
{
"surfaceUpdate": {
"surfaceId": "s1",
"components": [
{
"id": "weighted",
"weight": 2.0,
"component": {
"Column": {
"children": {"explicitList": ["a", "b"]},
},
},
},
],
},
},
{
"dataModelUpdate": {
"surfaceId": "s1",
"contents": [
{"key": "name", "valueString": "Alice"},
{"key": "score", "valueNumber": 42},
{"key": "active", "valueBoolean": True},
],
},
},
{
"dataModelUpdate": {
"surfaceId": "s1",
"path": "/user",
"contents": [
{
"key": "prefs",
"valueMap": [
{"key": "theme", "valueString": "dark"},
],
},
],
},
},
{
"deleteSurface": {"surfaceId": "s1"},
},
]
# ---------------------------------------------------------------------------
# Invalid server-to-client payloads
# ---------------------------------------------------------------------------
INVALID_SERVER_MESSAGES: list[dict[str, Any]] = [
{},
{"beginRendering": {"surfaceId": "s1"}},
{"surfaceUpdate": {"surfaceId": "s1", "components": []}},
{
"beginRendering": {"surfaceId": "s1", "root": "r"},
"deleteSurface": {"surfaceId": "s1"},
},
{"unknownType": {"surfaceId": "s1"}},
]
# ---------------------------------------------------------------------------
# Valid client-to-server payloads
# ---------------------------------------------------------------------------
VALID_CLIENT_EVENTS: list[dict[str, Any]] = [
{
"userAction": {
"name": "click",
"surfaceId": "s1",
"sourceComponentId": "btn-1",
"timestamp": "2026-03-12T10:00:00Z",
"context": {},
},
},
{
"userAction": {
"name": "submit",
"surfaceId": "s1",
"sourceComponentId": "btn-2",
"timestamp": "2026-03-12T10:00:00Z",
"context": {"field": "value"},
},
},
{
"error": {"message": "render failed", "code": 500},
},
]
# ---------------------------------------------------------------------------
# Invalid client-to-server payloads
# ---------------------------------------------------------------------------
INVALID_CLIENT_EVENTS: list[dict[str, Any]] = [
{},
{"userAction": {"name": "click"}},
{
"userAction": {
"name": "click",
"surfaceId": "s1",
"sourceComponentId": "btn-1",
"timestamp": "2026-03-12T10:00:00Z",
"context": {},
},
"error": {"message": "oops"},
},
]
# ---------------------------------------------------------------------------
# Catalog component payloads (validated structurally)
# ---------------------------------------------------------------------------
VALID_COMPONENTS: dict[str, dict[str, Any]] = {
"Text": {"text": {"literalString": "hello"}, "usageHint": "h1"},
"Image": {"url": {"path": "/img/url"}, "fit": "cover", "usageHint": "avatar"},
"Icon": {"name": {"literalString": "home"}},
"Video": {"url": {"literalString": "https://example.com/video.mp4"}},
"AudioPlayer": {"url": {"literalString": "https://example.com/audio.mp3"}},
"Row": {"children": {"explicitList": ["a", "b"]}, "distribution": "center"},
"Column": {"children": {"template": {"componentId": "c1", "dataBinding": "/list"}}},
"List": {"children": {"explicitList": ["x"]}, "direction": "horizontal"},
"Card": {"child": "inner"},
"Tabs": {"tabItems": [{"title": {"literalString": "Tab 1"}, "child": "content"}]},
"Divider": {"axis": "horizontal"},
"Modal": {"entryPointChild": "trigger", "contentChild": "body"},
"Button": {"child": "label", "action": {"name": "go"}},
"CheckBox": {"label": {"literalString": "Accept"}, "value": {"literalBoolean": False}},
"TextField": {"label": {"literalString": "Name"}},
"DateTimeInput": {"value": {"path": "/date"}},
"MultipleChoice": {
"selections": {"literalArray": ["a"]},
"options": [{"label": {"literalString": "A"}, "value": "a"}],
},
"Slider": {"value": {"literalNumber": 50}, "minValue": 0, "maxValue": 100},
}
class TestServerToClientConformance:
"""Pydantic models and JSON schema must agree on server-to-client messages."""
@pytest.mark.parametrize("payload", VALID_SERVER_MESSAGES)
def test_valid_accepted_by_both(self, payload: dict[str, Any]) -> None:
assert _json_schema_valid(SERVER_SCHEMA, payload), (
f"JSON schema rejected valid payload: {payload}"
)
assert _pydantic_valid_message(payload), (
f"Pydantic rejected valid payload: {payload}"
)
@pytest.mark.parametrize("payload", INVALID_SERVER_MESSAGES)
def test_invalid_rejected_by_pydantic(self, payload: dict[str, Any]) -> None:
assert not _pydantic_valid_message(payload), (
f"Pydantic accepted invalid payload: {payload}"
)
class TestClientToServerConformance:
"""Pydantic models and JSON schema must agree on client-to-server events."""
@pytest.mark.parametrize("payload", VALID_CLIENT_EVENTS)
def test_valid_accepted_by_both(self, payload: dict[str, Any]) -> None:
assert _json_schema_valid(CLIENT_SCHEMA, payload), (
f"JSON schema rejected valid payload: {payload}"
)
assert _pydantic_valid_event(payload), (
f"Pydantic rejected valid payload: {payload}"
)
@pytest.mark.parametrize("payload", INVALID_CLIENT_EVENTS)
def test_invalid_rejected_by_pydantic(self, payload: dict[str, Any]) -> None:
assert not _pydantic_valid_event(payload), (
f"Pydantic accepted invalid payload: {payload}"
)
class TestCatalogConformance:
"""Catalog component schemas and Pydantic models must define the same components."""
def test_catalog_component_names_match(self) -> None:
from crewai.a2a.extensions.a2ui.catalog import STANDARD_CATALOG_COMPONENTS
schema_components = set(CATALOG_SCHEMA["components"].keys())
assert schema_components == STANDARD_CATALOG_COMPONENTS
@pytest.mark.parametrize(
"name,props",
list(VALID_COMPONENTS.items()),
)
def test_valid_component_accepted_by_catalog_schema(
self, name: str, props: dict[str, Any]
) -> None:
component_schema = CATALOG_SCHEMA["components"][name]
assert _json_schema_valid(component_schema, props), (
f"Catalog schema rejected valid {name}: {props}"
)
@pytest.mark.parametrize(
"name,props",
list(VALID_COMPONENTS.items()),
)
def test_valid_component_accepted_by_pydantic(
self, name: str, props: dict[str, Any]
) -> None:
model_cls = getattr(catalog, name)
try:
model_cls.model_validate(props)
except Exception as exc:
pytest.fail(f"Pydantic {name} rejected valid props: {exc}")
def test_catalog_required_fields_match(self) -> None:
"""Required fields in the JSON schema match non-optional Pydantic fields."""
for comp_name, comp_schema in CATALOG_SCHEMA["components"].items():
schema_required = set(comp_schema.get("required", []))
model_cls = getattr(catalog, comp_name)
pydantic_required = {
info.alias or field_name
for field_name, info in model_cls.model_fields.items()
if info.is_required()
}
assert schema_required == pydantic_required, (
f"{comp_name}: schema requires {schema_required}, "
f"Pydantic requires {pydantic_required}"
)
def test_catalog_fields_match(self) -> None:
"""Field names in JSON schema match Pydantic model aliases."""
for comp_name, comp_schema in CATALOG_SCHEMA["components"].items():
schema_fields = set(comp_schema.get("properties", {}).keys())
model_cls = getattr(catalog, comp_name)
pydantic_fields = {
info.alias or field_name
for field_name, info in model_cls.model_fields.items()
}
assert schema_fields == pydantic_fields, (
f"{comp_name}: schema has {schema_fields}, "
f"Pydantic has {pydantic_fields}"
)

View File

@@ -1,109 +1,104 @@
interactions:
- request:
body: '{"messages":[{"role":"system","content":"You are Researcher. You''re love
to sey howdy.\nYour personal goal is: Be super empathetic."},{"role":"user","content":"\nCurrent
Task: say howdy\n\nThis is the expected criteria for your final answer: Howdy!\nyou
MUST return the actual complete content as the final answer, not a summary.\n\nProvide
your complete response:"}],"model":"gpt-4.1-mini"}'
body: '{"messages": [{"role": "system", "content": "You are Researcher. You''re
love to sey howdy.\nYour personal goal is: Be super empathetic.\nTo give my
best complete final answer to the task use 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 howdy\n\nThis is the expect criteria for your final answer: Howdy!\nyou
MUST return the actual complete content as the final answer, not a summary.\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"}'
headers:
User-Agent:
- X-USER-AGENT-XXX
accept:
- application/json
accept-encoding:
- ACCEPT-ENCODING-XXX
authorization:
- AUTHORIZATION-XXX
- gzip, deflate
connection:
- keep-alive
content-length:
- '391'
- '784'
content-type:
- application/json
cookie:
- __cf_bm=9.8sBYBkvBR8R1K_bVF7xgU..80XKlEIg3N2OBbTSCU-1727214102-1.0.1.1-.qiTLXbPamYUMSuyNsOEB9jhGu.jOifujOrx9E2JZvStbIZ9RTIiE44xKKNfLPxQkOi6qAT3h6htK8lPDGV_5g;
_cfuvid=lbRdAddVWV6W3f5Dm9SaOPWDUOxqtZBSPr_fTW26nEA-1727213194587-0.0.1.1-604800000
host:
- api.openai.com
user-agent:
- OpenAI/Python 1.47.0
x-stainless-arch:
- X-STAINLESS-ARCH-XXX
- arm64
x-stainless-async:
- 'false'
x-stainless-lang:
- python
x-stainless-os:
- X-STAINLESS-OS-XXX
- MacOS
x-stainless-package-version:
- 1.83.0
x-stainless-read-timeout:
- X-STAINLESS-READ-TIMEOUT-XXX
x-stainless-retry-count:
- '0'
- 1.47.0
x-stainless-raw-response:
- 'true'
x-stainless-runtime:
- CPython
x-stainless-runtime-version:
- 3.13.12
- 3.11.7
method: POST
uri: https://api.openai.com/v1/chat/completions
response:
body:
string: "{\n \"id\": \"chatcmpl-DJVU3yUdUSuW0vgTcxubm9K2Q2cT6\",\n \"object\":
\"chat.completion\",\n \"created\": 1773541627,\n \"model\": \"gpt-4.1-mini-2025-04-14\",\n
\ \"choices\": [\n {\n \"index\": 0,\n \"message\": {\n \"role\":
\"assistant\",\n \"content\": \"Howdy!\",\n \"refusal\": null,\n
\ \"annotations\": []\n },\n \"logprobs\": null,\n \"finish_reason\":
\"stop\"\n }\n ],\n \"usage\": {\n \"prompt_tokens\": 75,\n \"completion_tokens\":
2,\n \"total_tokens\": 77,\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_db58fd2815\"\n}\n"
string: "{\n \"id\": \"chatcmpl-AB7cCuywn5zE7q0S8IXWVnXoVE81Y\",\n \"object\"\
: \"chat.completion\",\n \"created\": 1727214244,\n \"model\": \"gpt-4o-2024-05-13\"\
,\n \"choices\": [\n {\n \"index\": 0,\n \"message\": {\n \
\ \"role\": \"assistant\",\n \"content\": \"I now can give a great\
\ answer \\nFinal Answer: Howdy!\",\n \"refusal\": null\n },\n\
\ \"logprobs\": null,\n \"finish_reason\": \"stop\"\n }\n ],\n\
\ \"usage\": {\n \"prompt_tokens\": 159,\n \"completion_tokens\": 14,\n\
\ \"total_tokens\": 173,\n \"completion_tokens_details\": {\n \"\
reasoning_tokens\": 0\n }\n },\n \"system_fingerprint\": \"fp_a2ff031fb5\"\
\n}\n"
headers:
CF-Cache-Status:
- DYNAMIC
CF-Ray:
- 9dc813a23bf323dd-EWR
CF-RAY:
- 8c85f41ffdb81cf3-GRU
Connection:
- keep-alive
Content-Type:
- application/json
Date:
- Sun, 15 Mar 2026 02:27:07 GMT
- Tue, 24 Sep 2024 21:44:04 GMT
Server:
- cloudflare
Strict-Transport-Security:
- STS-XXX
Transfer-Encoding:
- chunked
X-Content-Type-Options:
- X-CONTENT-TYPE-XXX
- nosniff
access-control-expose-headers:
- ACCESS-CONTROL-XXX
alt-svc:
- h3=":443"; ma=86400
- X-Request-ID
openai-organization:
- OPENAI-ORG-XXX
- crewai-iuxna1
openai-processing-ms:
- '366'
openai-project:
- OPENAI-PROJECT-XXX
- '243'
openai-version:
- '2020-10-01'
set-cookie:
- SET-COOKIE-XXX
x-openai-proxy-wasm:
- v0.1
strict-transport-security:
- max-age=31536000; includeSubDomains; preload
x-ratelimit-limit-requests:
- X-RATELIMIT-LIMIT-REQUESTS-XXX
- '10000'
x-ratelimit-limit-tokens:
- X-RATELIMIT-LIMIT-TOKENS-XXX
- '30000000'
x-ratelimit-remaining-requests:
- X-RATELIMIT-REMAINING-REQUESTS-XXX
- '9999'
x-ratelimit-remaining-tokens:
- X-RATELIMIT-REMAINING-TOKENS-XXX
- '29999815'
x-ratelimit-reset-requests:
- X-RATELIMIT-RESET-REQUESTS-XXX
- 6ms
x-ratelimit-reset-tokens:
- X-RATELIMIT-RESET-TOKENS-XXX
- 0s
x-request-id:
- X-REQUEST-ID-XXX
- req_50ed3333fd70ce8e32abd43dbe7f9362
status:
code: 200
message: OK

View File

@@ -39,13 +39,13 @@ interactions:
x-stainless-runtime:
- CPython
x-stainless-runtime-version:
- 3.13.12
- 3.13.3
method: POST
uri: https://api.openai.com/v1/chat/completions
response:
body:
string: "{\n \"id\": \"chatcmpl-DJVXy6jcneOpe2GdSSGfNO4TkB6np\",\n \"object\":
\"chat.completion\",\n \"created\": 1773541870,\n \"model\": \"gpt-4.1-mini-2025-04-14\",\n
string: "{\n \"id\": \"chatcmpl-DIqrxbdWncBetSyqX8P36UUXoil9d\",\n \"object\":
\"chat.completion\",\n \"created\": 1773385505,\n \"model\": \"gpt-4.1-mini-2025-04-14\",\n
\ \"choices\": [\n {\n \"index\": 0,\n \"message\": {\n \"role\":
\"assistant\",\n \"content\": \"Test expected output\",\n \"refusal\":
null,\n \"annotations\": []\n },\n \"logprobs\": null,\n
@@ -54,18 +54,18 @@ interactions:
{\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_db58fd2815\"\n}\n"
\"default\",\n \"system_fingerprint\": \"fp_5e793402c9\"\n}\n"
headers:
CF-Cache-Status:
- DYNAMIC
CF-Ray:
- 9dc819af5cbb8ce8-EWR
- 9db9302f7f411efc-EWR
Connection:
- keep-alive
Content-Type:
- application/json
Date:
- Sun, 15 Mar 2026 02:31:10 GMT
- Fri, 13 Mar 2026 07:05:06 GMT
Server:
- cloudflare
Strict-Transport-Security:
@@ -81,7 +81,7 @@ interactions:
openai-organization:
- OPENAI-ORG-XXX
openai-processing-ms:
- '351'
- '376'
openai-project:
- OPENAI-PROJECT-XXX
openai-version:

View File

@@ -7,22 +7,22 @@ interactions:
Task: Produce and amazing 1 paragraph draft of an article about AI Agents.\n\nThis
is the expected criteria for your final answer: A 4 paragraph article about
AI.\nyou MUST return the actual complete content as the final answer, not a
summary."}],"model":"gpt-4.1-mini","tool_choice":"auto","tools":[{"type":"function","function":{"name":"delegate_work_to_coworker","description":"Delegate
summary.\n\nThis is VERY important to you, your job depends on it!"}],"model":"gpt-4.1-mini","tool_choice":"auto","tools":[{"type":"function","function":{"name":"Delegate_work_to_coworker","description":"Delegate
a specific task to one of the following coworkers: Senior Writer\nThe input
to this tool should be the coworker, the task you want them to do, and ALL necessary
context to execute the task, they know nothing about the task, so share absolutely
everything you know, don''t reference things but instead explain them.","strict":true,"parameters":{"properties":{"task":{"description":"The
everything you know, don''t reference things but instead explain them.","parameters":{"properties":{"task":{"description":"The
task to delegate","title":"Task","type":"string"},"context":{"description":"The
context for the task","title":"Context","type":"string"},"coworker":{"description":"The
role/name of the coworker to delegate to","title":"Coworker","type":"string"}},"required":["task","context","coworker"],"type":"object","additionalProperties":false}}},{"type":"function","function":{"name":"ask_question_to_coworker","description":"Ask
role/name of the coworker to delegate to","title":"Coworker","type":"string"}},"required":["task","context","coworker"],"type":"object"}}},{"type":"function","function":{"name":"Ask_question_to_coworker","description":"Ask
a specific question to one of the following coworkers: Senior Writer\nThe input
to this tool should be the coworker, the question you have for them, and ALL
necessary context to ask the question properly, they know nothing about the
question, so share absolutely everything you know, don''t reference things but
instead explain them.","strict":true,"parameters":{"properties":{"question":{"description":"The
instead explain them.","parameters":{"properties":{"question":{"description":"The
question to ask","title":"Question","type":"string"},"context":{"description":"The
context for the question","title":"Context","type":"string"},"coworker":{"description":"The
role/name of the coworker to ask","title":"Coworker","type":"string"}},"required":["question","context","coworker"],"type":"object","additionalProperties":false}}}]}'
role/name of the coworker to ask","title":"Coworker","type":"string"}},"required":["question","context","coworker"],"type":"object"}}}]}'
headers:
User-Agent:
- X-USER-AGENT-XXX
@@ -35,7 +35,7 @@ interactions:
connection:
- keep-alive
content-length:
- '2298'
- '2270'
content-type:
- application/json
host:
@@ -57,46 +57,48 @@ interactions:
x-stainless-runtime:
- CPython
x-stainless-runtime-version:
- 3.13.12
- 3.13.3
method: POST
uri: https://api.openai.com/v1/chat/completions
response:
body:
string: "{\n \"id\": \"chatcmpl-DJVUF1zvo6c1aeolU4hCgOK2XeUEK\",\n \"object\":
\"chat.completion\",\n \"created\": 1773541639,\n \"model\": \"gpt-4.1-mini-2025-04-14\",\n
string: "{\n \"id\": \"chatcmpl-D0uJLGWT3ELLQ84FugHD30N0rap1d\",\n \"object\":
\"chat.completion\",\n \"created\": 1769108831,\n \"model\": \"gpt-4.1-mini-2025-04-14\",\n
\ \"choices\": [\n {\n \"index\": 0,\n \"message\": {\n \"role\":
\"assistant\",\n \"content\": null,\n \"tool_calls\": [\n {\n
\ \"id\": \"call_wi7jZ8PKxWMadJufpDtwfLF3\",\n \"type\":
\"function\",\n \"function\": {\n \"name\": \"delegate_work_to_coworker\",\n
\ \"arguments\": \"{\\\"task\\\":\\\"Produce an amazing 1 paragraph
draft of an article about AI Agents. This paragraph should be engaging, informative,
and provide a clear introduction to the topic of AI Agents, capturing the
readers' interest and setting the stage for a longer article.\\\",\\\"context\\\":\\\"The
article on AI Agents will be a 4 paragraph piece aiming to explain what AI
Agents are, their functionalities, importance, and potential future developments.
The draft paragraph should serve as a compelling introduction that hooks the
audience and introduces the subject matter effectively.\\\",\\\"coworker\\\":\\\"Senior
Writer\\\"}\"\n }\n }\n ],\n \"refusal\":
null,\n \"annotations\": []\n },\n \"logprobs\": null,\n
\ \"finish_reason\": \"tool_calls\"\n }\n ],\n \"usage\": {\n \"prompt_tokens\":
386,\n \"completion_tokens\": 124,\n \"total_tokens\": 510,\n \"prompt_tokens_details\":
{\n \"cached_tokens\": 0,\n \"audio_tokens\": 0\n },\n \"completion_tokens_details\":
\ \"id\": \"call_j5vDsg6M6N1UbDUrQTZnwKia\",\n \"type\":
\"function\",\n \"function\": {\n \"name\": \"Delegate_work_to_coworker\",\n
\ \"arguments\": \"{\\\"coworker\\\":\\\"Senior Writer\\\",\\\"task\\\":\\\"Produce
a 4-paragraph article about AI focusing on AI Agents, starting with a 1-paragraph
draft. The article should be engaging, informative, and demonstrate a deep
understanding of the topic, highlighting how AI Agents function, their applications,
benefits and potential future developments.\\\",\\\"context\\\":\\\"The task
is to produce an amazing 4-paragraph article about AI with a focus on AI Agents.
The first deliverable is a 1-paragraph draft that sets the tone for the full
article, emphasizing the significance and capabilities of AI Agents in modern
technology. The final article should be coherent, seamlessly cover the topic,
and be suitable for publication on a technology-focused platform.\\\"}\"\n
\ }\n }\n ],\n \"refusal\": null,\n \"annotations\":
[]\n },\n \"logprobs\": null,\n \"finish_reason\": \"tool_calls\"\n
\ }\n ],\n \"usage\": {\n \"prompt_tokens\": 399,\n \"completion_tokens\":
159,\n \"total_tokens\": 558,\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_5e793402c9\"\n}\n"
\"default\",\n \"system_fingerprint\": \"fp_376a7ccef1\"\n}\n"
headers:
CF-Cache-Status:
- DYNAMIC
CF-Ray:
- 9dc8140db9f85f83-EWR
CF-RAY:
- CF-RAY-XXX
Connection:
- keep-alive
Content-Type:
- application/json
Date:
- Sun, 15 Mar 2026 02:27:20 GMT
- Thu, 22 Jan 2026 19:07:14 GMT
Server:
- cloudflare
Set-Cookie:
- SET-COOKIE-XXX
Strict-Transport-Security:
- STS-XXX
Transfer-Encoding:
@@ -107,16 +109,18 @@ interactions:
- ACCESS-CONTROL-XXX
alt-svc:
- h3=":443"; ma=86400
cf-cache-status:
- DYNAMIC
openai-organization:
- OPENAI-ORG-XXX
openai-processing-ms:
- '1331'
- '2967'
openai-project:
- OPENAI-PROJECT-XXX
openai-version:
- '2020-10-01'
set-cookie:
- SET-COOKIE-XXX
x-envoy-upstream-service-time:
- '2989'
x-openai-proxy-wasm:
- v0.1
x-ratelimit-limit-requests:
@@ -140,18 +144,25 @@ interactions:
body: '{"messages":[{"role":"system","content":"You are Senior Writer. You''re
a senior writer, specialized in technology, software engineering, AI and startups.
You work as a freelancer and are now working on writing content for a new customer.\nYour
personal goal is: Write the best content about AI and AI agents."},{"role":"user","content":"\nCurrent
Task: Produce an amazing 1 paragraph draft of an article about AI Agents. This
paragraph should be engaging, informative, and provide a clear introduction
to the topic of AI Agents, capturing the readers'' interest and setting the
stage for a longer article.\n\nThis is the expected criteria for your final
answer: Your best answer to your coworker asking you this, accounting for the
context shared.\nyou MUST return the actual complete content as the final answer,
not a summary.\n\nThis is the context you''re working with:\nThe article on
AI Agents will be a 4 paragraph piece aiming to explain what AI Agents are,
their functionalities, importance, and potential future developments. The draft
paragraph should serve as a compelling introduction that hooks the audience
and introduces the subject matter effectively.\n\nProvide your complete response:"}],"model":"gpt-4.1-mini"}'
personal goal is: Write the best content about AI and AI agents.\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:
Produce a 4-paragraph article about AI focusing on AI Agents, starting with
a 1-paragraph draft. The article should be engaging, informative, and demonstrate
a deep understanding of the topic, highlighting how AI Agents function, their
applications, benefits and potential future developments.\n\nThis is the expected
criteria for your final answer: Your best answer to your coworker asking you
this, accounting for the context shared.\nyou MUST return the actual complete
content as the final answer, not a summary.\n\nThis is the context you''re working
with:\nThe task is to produce an amazing 4-paragraph article about AI with a
focus on AI Agents. The first deliverable is a 1-paragraph draft that sets the
tone for the full article, emphasizing the significance and capabilities of
AI Agents in modern technology. The final article should be coherent, seamlessly
cover the topic, and be suitable for publication on a technology-focused platform.\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-4.1-mini"}'
headers:
User-Agent:
- X-USER-AGENT-XXX
@@ -164,7 +175,7 @@ interactions:
connection:
- keep-alive
content-length:
- '1228'
- '1766'
content-type:
- application/json
host:
@@ -186,46 +197,53 @@ interactions:
x-stainless-runtime:
- CPython
x-stainless-runtime-version:
- 3.13.12
- 3.13.3
method: POST
uri: https://api.openai.com/v1/chat/completions
response:
body:
string: "{\n \"id\": \"chatcmpl-DJVUHEhPlZEKRnV2LXW33Z2y5jZB6\",\n \"object\":
\"chat.completion\",\n \"created\": 1773541641,\n \"model\": \"gpt-4.1-mini-2025-04-14\",\n
string: "{\n \"id\": \"chatcmpl-D0uJOvgyklVDFe1l4S0ty6oMuhfak\",\n \"object\":
\"chat.completion\",\n \"created\": 1769108834,\n \"model\": \"gpt-4.1-mini-2025-04-14\",\n
\ \"choices\": [\n {\n \"index\": 0,\n \"message\": {\n \"role\":
\"assistant\",\n \"content\": \"In the rapidly evolving landscape of
artificial intelligence, AI agents stand out as transformative entities capable
of perceiving their environment, making decisions, and autonomously executing
tasks to achieve specific goals. Far beyond simple algorithms, these agents
act as intelligent intermediaries that can range from virtual assistants managing
our schedules to complex systems driving autonomous vehicles or optimizing
supply chains. As AI agents increasingly integrate into daily life and business
operations, understanding their core principles and potential impact is essential\u2014not
just for tech enthusiasts but for anyone interested in how intelligent automation
is reshaping our future. This article will explore what AI agents are, how
they function, their growing significance across industries, and the exciting
possibilities they hold on the horizon.\",\n \"refusal\": null,\n \"annotations\":
\"assistant\",\n \"content\": \"Thought: I now can give a great answer\\nFinal
Answer: \\n\\nArtificial Intelligence (AI) agents have emerged as a transformative
force in modern technology, redefining how machines interact with the world
and assist humans in complex tasks. These autonomous systems are designed
to perceive their environment, make decisions, and execute actions to achieve
specific goals, often adapting in real-time to changing conditions. Unlike
traditional software that follows pre-programmed instructions without deviation,
AI agents exhibit a level of intelligence akin to decision-making entities,
leveraging advanced algorithms in machine learning, natural language processing,
and computer vision. From virtual assistants like Siri and Alexa to sophisticated
industrial robots and predictive analytics systems, AI agents are not only
enhancing efficiency but also opening new frontiers in automation and human-computer
interaction. Their capacity to learn, reason, and self-improve positions them
as pivotal enablers in sectors ranging from healthcare and finance to autonomous
vehicles and smart cities, heralding a future where intelligent agents will
seamlessly augment daily life and enterprise operations. This article delves
into the inner workings of AI agents, explores their diverse applications,
underscores their benefits, and envisions their evolving role in shaping the
technological landscape.\",\n \"refusal\": null,\n \"annotations\":
[]\n },\n \"logprobs\": null,\n \"finish_reason\": \"stop\"\n
\ }\n ],\n \"usage\": {\n \"prompt_tokens\": 222,\n \"completion_tokens\":
137,\n \"total_tokens\": 359,\n \"prompt_tokens_details\": {\n \"cached_tokens\":
\ }\n ],\n \"usage\": {\n \"prompt_tokens\": 340,\n \"completion_tokens\":
233,\n \"total_tokens\": 573,\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_5e793402c9\"\n}\n"
\"default\",\n \"system_fingerprint\": \"fp_2191215734\"\n}\n"
headers:
CF-Cache-Status:
- DYNAMIC
CF-Ray:
- 9dc814182b1c281b-EWR
CF-RAY:
- CF-RAY-XXX
Connection:
- keep-alive
Content-Type:
- application/json
Date:
- Sun, 15 Mar 2026 02:27:22 GMT
- Thu, 22 Jan 2026 19:07:18 GMT
Server:
- cloudflare
Set-Cookie:
- SET-COOKIE-XXX
Strict-Transport-Security:
- STS-XXX
Transfer-Encoding:
@@ -236,16 +254,18 @@ interactions:
- ACCESS-CONTROL-XXX
alt-svc:
- h3=":443"; ma=86400
cf-cache-status:
- DYNAMIC
openai-organization:
- OPENAI-ORG-XXX
openai-processing-ms:
- '1582'
- '3760'
openai-project:
- OPENAI-PROJECT-XXX
openai-version:
- '2020-10-01'
set-cookie:
- SET-COOKIE-XXX
x-envoy-upstream-service-time:
- '3776'
x-openai-proxy-wasm:
- v0.1
x-ratelimit-limit-requests:
@@ -266,50 +286,57 @@ interactions:
code: 200
message: OK
- request:
body: "{\"messages\":[{\"role\":\"system\",\"content\":\"You are CEO. You're an
long time CEO of a content creation agency with a Senior Writer on the team.
You're now working on a new project and want to make sure the content produced
is amazing.\\nYour personal goal is: Make sure the writers in your company produce
amazing content.\"},{\"role\":\"user\",\"content\":\"\\nCurrent Task: Produce
and amazing 1 paragraph draft of an article about AI Agents.\\n\\nThis is the
expected criteria for your final answer: A 4 paragraph article about AI.\\nyou
MUST return the actual complete content as the final answer, not a summary.\"},{\"role\":\"assistant\",\"content\":null,\"tool_calls\":[{\"id\":\"call_wi7jZ8PKxWMadJufpDtwfLF3\",\"type\":\"function\",\"function\":{\"name\":\"delegate_work_to_coworker\",\"arguments\":\"{\\\"task\\\":\\\"Produce
an amazing 1 paragraph draft of an article about AI Agents. This paragraph should
be engaging, informative, and provide a clear introduction to the topic of AI
Agents, capturing the readers' interest and setting the stage for a longer article.\\\",\\\"context\\\":\\\"The
article on AI Agents will be a 4 paragraph piece aiming to explain what AI Agents
are, their functionalities, importance, and potential future developments. The
draft paragraph should serve as a compelling introduction that hooks the audience
and introduces the subject matter effectively.\\\",\\\"coworker\\\":\\\"Senior
Writer\\\"}\"}}]},{\"role\":\"tool\",\"tool_call_id\":\"call_wi7jZ8PKxWMadJufpDtwfLF3\",\"name\":\"delegate_work_to_coworker\",\"content\":\"In
the rapidly evolving landscape of artificial intelligence, AI agents stand out
as transformative entities capable of perceiving their environment, making decisions,
and autonomously executing tasks to achieve specific goals. Far beyond simple
algorithms, these agents act as intelligent intermediaries that can range from
virtual assistants managing our schedules to complex systems driving autonomous
vehicles or optimizing supply chains. As AI agents increasingly integrate into
daily life and business operations, understanding their core principles and
potential impact is essential\u2014not just for tech enthusiasts but for anyone
interested in how intelligent automation is reshaping our future. This article
will explore what AI agents are, how they function, their growing significance
across industries, and the exciting possibilities they hold on the horizon.\"},{\"role\":\"user\",\"content\":\"Analyze
body: '{"messages":[{"role":"system","content":"You are CEO. You''re an long time
CEO of a content creation agency with a Senior Writer on the team. You''re now
working on a new project and want to make sure the content produced is amazing.\nYour
personal goal is: Make sure the writers in your company produce amazing content."},{"role":"user","content":"\nCurrent
Task: Produce and amazing 1 paragraph draft of an article about AI Agents.\n\nThis
is the expected criteria for your final answer: A 4 paragraph article about
AI.\nyou MUST return the actual complete content as the final answer, not a
summary.\n\nThis is VERY important to you, your job depends on it!"},{"role":"assistant","content":null,"tool_calls":[{"id":"call_j5vDsg6M6N1UbDUrQTZnwKia","type":"function","function":{"name":"Delegate_work_to_coworker","arguments":"{\"coworker\":\"Senior
Writer\",\"task\":\"Produce a 4-paragraph article about AI focusing on AI Agents,
starting with a 1-paragraph draft. The article should be engaging, informative,
and demonstrate a deep understanding of the topic, highlighting how AI Agents
function, their applications, benefits and potential future developments.\",\"context\":\"The
task is to produce an amazing 4-paragraph article about AI with a focus on AI
Agents. The first deliverable is a 1-paragraph draft that sets the tone for
the full article, emphasizing the significance and capabilities of AI Agents
in modern technology. The final article should be coherent, seamlessly cover
the topic, and be suitable for publication on a technology-focused platform.\"}"}}]},{"role":"tool","tool_call_id":"call_j5vDsg6M6N1UbDUrQTZnwKia","content":"Artificial
Intelligence (AI) agents have emerged as a transformative force in modern technology,
redefining how machines interact with the world and assist humans in complex
tasks. These autonomous systems are designed to perceive their environment,
make decisions, and execute actions to achieve specific goals, often adapting
in real-time to changing conditions. Unlike traditional software that follows
pre-programmed instructions without deviation, AI agents exhibit a level of
intelligence akin to decision-making entities, leveraging advanced algorithms
in machine learning, natural language processing, and computer vision. From
virtual assistants like Siri and Alexa to sophisticated industrial robots and
predictive analytics systems, AI agents are not only enhancing efficiency but
also opening new frontiers in automation and human-computer interaction. Their
capacity to learn, reason, and self-improve positions them as pivotal enablers
in sectors ranging from healthcare and finance to autonomous vehicles and smart
cities, heralding a future where intelligent agents will seamlessly augment
daily life and enterprise operations. This article delves into the inner workings
of AI agents, explores their diverse applications, underscores their benefits,
and envisions their evolving role in shaping the technological landscape."},{"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-4.1-mini\",\"tool_choice\":\"auto\",\"tools\":[{\"type\":\"function\",\"function\":{\"name\":\"delegate_work_to_coworker\",\"description\":\"Delegate
a specific task to one of the following coworkers: Senior Writer\\nThe input
call the next tool. Deliver only the answer without meta-commentary."}],"model":"gpt-4.1-mini","tool_choice":"auto","tools":[{"type":"function","function":{"name":"Delegate_work_to_coworker","description":"Delegate
a specific task to one of the following coworkers: Senior Writer\nThe input
to this tool should be the coworker, the task you want them to do, and ALL necessary
context to execute the task, they know nothing about the task, so share absolutely
everything you know, don't reference things but instead explain them.\",\"strict\":true,\"parameters\":{\"properties\":{\"task\":{\"description\":\"The
task to delegate\",\"title\":\"Task\",\"type\":\"string\"},\"context\":{\"description\":\"The
context for the task\",\"title\":\"Context\",\"type\":\"string\"},\"coworker\":{\"description\":\"The
role/name of the coworker to delegate to\",\"title\":\"Coworker\",\"type\":\"string\"}},\"required\":[\"task\",\"context\",\"coworker\"],\"type\":\"object\",\"additionalProperties\":false}}},{\"type\":\"function\",\"function\":{\"name\":\"ask_question_to_coworker\",\"description\":\"Ask
a specific question to one of the following coworkers: Senior Writer\\nThe input
everything you know, don''t reference things but instead explain them.","parameters":{"properties":{"task":{"description":"The
task to delegate","title":"Task","type":"string"},"context":{"description":"The
context for the task","title":"Context","type":"string"},"coworker":{"description":"The
role/name of the coworker to delegate to","title":"Coworker","type":"string"}},"required":["task","context","coworker"],"type":"object"}}},{"type":"function","function":{"name":"Ask_question_to_coworker","description":"Ask
a specific question to one of the following coworkers: Senior Writer\nThe input
to this tool should be the coworker, the question you have for them, and ALL
necessary context to ask the question properly, they know nothing about the
question, so share absolutely everything you know, don't reference things but
instead explain them.\",\"strict\":true,\"parameters\":{\"properties\":{\"question\":{\"description\":\"The
question to ask\",\"title\":\"Question\",\"type\":\"string\"},\"context\":{\"description\":\"The
context for the question\",\"title\":\"Context\",\"type\":\"string\"},\"coworker\":{\"description\":\"The
role/name of the coworker to ask\",\"title\":\"Coworker\",\"type\":\"string\"}},\"required\":[\"question\",\"context\",\"coworker\"],\"type\":\"object\",\"additionalProperties\":false}}}]}"
question, so share absolutely everything you know, don''t reference things but
instead explain them.","parameters":{"properties":{"question":{"description":"The
question to ask","title":"Question","type":"string"},"context":{"description":"The
context for the question","title":"Context","type":"string"},"coworker":{"description":"The
role/name of the coworker to ask","title":"Coworker","type":"string"}},"required":["question","context","coworker"],"type":"object"}}}]}'
headers:
User-Agent:
- X-USER-AGENT-XXX
@@ -322,7 +349,7 @@ interactions:
connection:
- keep-alive
content-length:
- '4241'
- '4785'
content-type:
- application/json
cookie:
@@ -346,67 +373,81 @@ interactions:
x-stainless-runtime:
- CPython
x-stainless-runtime-version:
- 3.13.12
- 3.13.3
method: POST
uri: https://api.openai.com/v1/chat/completions
response:
body:
string: "{\n \"id\": \"chatcmpl-DJVUIWVgzE4ZVBGyKaXO2HDAhJvSL\",\n \"object\":
\"chat.completion\",\n \"created\": 1773541642,\n \"model\": \"gpt-4.1-mini-2025-04-14\",\n
string: "{\n \"id\": \"chatcmpl-D0uJShynrNj5AQsUdLGMcR3mi57nv\",\n \"object\":
\"chat.completion\",\n \"created\": 1769108838,\n \"model\": \"gpt-4.1-mini-2025-04-14\",\n
\ \"choices\": [\n {\n \"index\": 0,\n \"message\": {\n \"role\":
\"assistant\",\n \"content\": \"AI Agents: Transforming the Future
with Intelligent Automation\\n\\nIn the rapidly evolving landscape of artificial
intelligence, AI agents stand out as transformative entities capable of perceiving
their environment, making decisions, and autonomously executing tasks to achieve
specific goals. Far beyond simple algorithms, these agents act as intelligent
intermediaries that can range from virtual assistants managing our schedules
to complex systems driving autonomous vehicles or optimizing supply chains.
As AI agents increasingly integrate into daily life and business operations,
understanding their core principles and potential impact is essential\u2014not
just for tech enthusiasts but for anyone interested in how intelligent automation
is reshaping our future. This article will explore what AI agents are, how
they function, their growing significance across industries, and the exciting
possibilities they hold on the horizon.\\n\\nAI agents operate by utilizing
a combination of machine learning, natural language processing, and decision-making
algorithms to interpret data, recognize patterns, and perform actions based
on their programming and learning experiences. These intelligent entities
can adapt to new information, improve their performance over time, and interact
with humans or other systems in a meaningful way. From customer service chatbots
that provide personalized support to AI-powered diagnostics in healthcare,
the versatility of AI agents highlights their capacity to enhance efficiency
and innovation across countless fields.\\n\\nThe importance of AI agents extends
beyond automation. They hold the promise of solving complex problems, augmenting
human capabilities, and making real-time decisions in environments where speed
and accuracy are critical. Industries such as finance, logistics, manufacturing,
and entertainment are experiencing disruptive changes driven by AI agents
that help optimize resources, reduce costs, and create new value propositions.
Moreover, ethical considerations and responsible development practices are
crucial to ensure these technologies benefit society while minimizing risks
and biases.\\n\\nLooking ahead, the future of AI agents is brimming with potential.
Advances in explainability, generalization, and collaboration between multiple
agents could lead to even more sophisticated systems capable of tackling challenges
that were once thought to be exclusively human domains. As AI agents continue
to evolve, their integration into everyday life will deepen, transforming
how we work, communicate, and solve problems\u2014ushering in a new era of
intelligent technology that empowers individuals and organizations alike.\",\n
\ \"refusal\": null,\n \"annotations\": []\n },\n \"logprobs\":
null,\n \"finish_reason\": \"stop\"\n }\n ],\n \"usage\": {\n \"prompt_tokens\":
695,\n \"completion_tokens\": 428,\n \"total_tokens\": 1123,\n \"prompt_tokens_details\":
{\n \"cached_tokens\": 0,\n \"audio_tokens\": 0\n },\n \"completion_tokens_details\":
\"assistant\",\n \"content\": \"Artificial Intelligence (AI) agents
have emerged as a transformative force in modern technology, redefining how
machines interact with the world and assist humans in complex tasks. These
autonomous systems are designed to perceive their environment, make decisions,
and execute actions to achieve specific goals, often adapting in real-time
to changing conditions. Unlike traditional software that follows pre-programmed
instructions without deviation, AI agents exhibit a level of intelligence
akin to decision-making entities, leveraging advanced algorithms in machine
learning, natural language processing, and computer vision. From virtual assistants
like Siri and Alexa to sophisticated industrial robots and predictive analytics
systems, AI agents are not only enhancing efficiency but also opening new
frontiers in automation and human-computer interaction. Their capacity to
learn, reason, and self-improve positions them as pivotal enablers in sectors
ranging from healthcare and finance to autonomous vehicles and smart cities,
heralding a future where intelligent agents will seamlessly augment daily
life and enterprise operations. This article delves into the inner workings
of AI agents, explores their diverse applications, underscores their benefits,
and envisions their evolving role in shaping the technological landscape.
\\n\\nAt the core, AI agents function by integrating sensory data with algorithms
that mimic human cognition, enabling them to interpret complex inputs and
make autonomous decisions. These agents operate through a cycle of perception,
reasoning, and action: perceiving their environment via sensors or input data;
processing this information through models that predict outcomes and strategize
moves; and finally, executing actions that influence or interact with the
external world. Machine learning plays a critical role, allowing agents to
improve their performance based on experience without explicit reprogramming.
Reinforcement learning, a subset of machine learning, teaches agents to learn
optimal behaviors by rewarding desirable outcomes. This intelligent adaptability
makes AI agents valuable in dynamic and unpredictable environments, where
rules and conditions continuously evolve.\\n\\nThe applications of AI agents
are vast and growing rapidly, spanning various industries and daily life realms.
In healthcare, AI agents support diagnostics, personalized treatment plans,
and patient monitoring, aiding doctors with data-driven insights. In finance,
they power algorithmic trading, fraud detection, and customer service chatbots.
Autonomous vehicles rely heavily on AI agents to navigate, interpret traffic
signals, and ensure passenger safety. Smart homes use these agents for energy
management and security, while industries deploy them in robotics for assembly
lines and quality control. These agents' ability to automate routine and complex
tasks leads to significant cost savings, higher precision, and scalability,
improving overall productivity and opening innovative business models.\\n\\nLooking
ahead, the evolution of AI agents promises even more profound impacts, with
advances in explainability, ethics, and collaboration between human and machine
intelligence. Future agents will likely be more transparent, offering clearer
reasoning for their decisions, which builds trust and accountability. Enhanced
multi-agent systems could coordinate complex tasks by sharing knowledge and
collaborating seamlessly. Moreover, ongoing research aims to ensure ethical
considerations are embedded from the ground up, addressing biases and safeguarding
privacy. As AI agents become increasingly integrated into society, they will
not only augment human abilities but also inspire new forms of creativity,
problem-solving, and interaction, ultimately shaping a more intelligent and
adaptive world.\",\n \"refusal\": null,\n \"annotations\": []\n
\ },\n \"logprobs\": null,\n \"finish_reason\": \"stop\"\n }\n
\ ],\n \"usage\": {\n \"prompt_tokens\": 825,\n \"completion_tokens\":
625,\n \"total_tokens\": 1450,\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_5e793402c9\"\n}\n"
\"default\",\n \"system_fingerprint\": \"fp_376a7ccef1\"\n}\n"
headers:
CF-Cache-Status:
- DYNAMIC
CF-Ray:
- 9dc814239fb65f83-EWR
CF-RAY:
- CF-RAY-XXX
Connection:
- keep-alive
Content-Type:
- application/json
Date:
- Sun, 15 Mar 2026 02:27:28 GMT
- Thu, 22 Jan 2026 19:07:28 GMT
Server:
- cloudflare
Strict-Transport-Security:
@@ -419,14 +460,18 @@ interactions:
- ACCESS-CONTROL-XXX
alt-svc:
- h3=":443"; ma=86400
cf-cache-status:
- DYNAMIC
openai-organization:
- OPENAI-ORG-XXX
openai-processing-ms:
- '4942'
- '9924'
openai-project:
- OPENAI-PROJECT-XXX
openai-version:
- '2020-10-01'
x-envoy-upstream-service-time:
- '9940'
x-openai-proxy-wasm:
- v0.1
x-ratelimit-limit-requests:

File diff suppressed because one or more lines are too long

View File

@@ -1,123 +1,247 @@
interactions:
- request:
body: '{"messages":[{"role":"system","content":"You are Researcher. You''re an
expert researcher, specialized in technology, software engineering, AI and startups.
You work as a freelancer and is now working on doing research and analysis for
a new customer.\nYour personal goal is: Make the best research and analysis
on content about AI and AI agents"},{"role":"user","content":"\nCurrent Task:
Write a test task\n\nThis is the expected criteria for your final answer: Test
output\nyou MUST return the actual complete content as the final answer, not
a summary."}],"model":"gpt-4.1-mini","tool_choice":"auto","tools":[{"type":"function","function":{"name":"another_test_tool","description":"Another
test tool","strict":true,"parameters":{"properties":{"query":{"description":"Query
to process","title":"Query","type":"string"}},"required":["query"],"type":"object","additionalProperties":false}}}]}'
body: '{"messages": [{"role": "system", "content": "You are Researcher. You''re
an expert researcher, specialized in technology, software engineering, AI and
startups. You work as a freelancer and is now working on doing research and
analysis for a new customer.\nYour personal goal is: Make the best research
and analysis on content about AI and AI agents\nYou ONLY have access to the
following tools, and should NEVER make up tools that are not listed here:\n\nTool
Name: Another Test Tool\nTool Arguments: {''query'': {''description'': ''Query
to process'', ''type'': ''str''}}\nTool Description: Another test tool\n\nUse
the following format:\n\nThought: you should always think about what to do\nAction:
the action to take, only one name of [Another Test Tool], just the name, exactly
as it''s written.\nAction Input: the input to the action, just a simple python
dictionary, enclosed in curly braces, using \" to wrap keys and values.\nObservation:
the result of the action\n\nOnce all necessary information is gathered:\n\nThought:
I now know the final answer\nFinal Answer: the final answer to the original
input question"}, {"role": "user", "content": "\nCurrent Task: Write a test
task\n\nThis is the expect criteria for your final answer: Test output\nyou
MUST return the actual complete content as the final answer, not a summary.\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", "stop": ["\nObservation:"],
"stream": false}'
headers:
User-Agent:
- X-USER-AGENT-XXX
accept:
- application/json
accept-encoding:
- ACCEPT-ENCODING-XXX
authorization:
- AUTHORIZATION-XXX
- gzip, deflate
connection:
- keep-alive
content-length:
- '892'
- '1525'
content-type:
- application/json
cookie:
- _cfuvid=eQzzWvIXDS8Me1OIBdCG5F1qFyVfAo3sumvYRE7J41E-1734965710778-0.0.1.1-604800000
host:
- api.openai.com
user-agent:
- OpenAI/Python 1.52.1
x-stainless-arch:
- X-STAINLESS-ARCH-XXX
- arm64
x-stainless-async:
- 'false'
x-stainless-lang:
- python
x-stainless-os:
- X-STAINLESS-OS-XXX
- MacOS
x-stainless-package-version:
- 1.83.0
x-stainless-read-timeout:
- X-STAINLESS-READ-TIMEOUT-XXX
- 1.52.1
x-stainless-raw-response:
- 'true'
x-stainless-retry-count:
- '0'
x-stainless-runtime:
- CPython
x-stainless-runtime-version:
- 3.13.12
- 3.12.7
method: POST
uri: https://api.openai.com/v1/chat/completions
response:
body:
string: "{\n \"id\": \"chatcmpl-DJVYArMj7gxUpliUsjoQOzDxaw0Ta\",\n \"object\":
\"chat.completion\",\n \"created\": 1773541882,\n \"model\": \"gpt-4.1-mini-2025-04-14\",\n
\ \"choices\": [\n {\n \"index\": 0,\n \"message\": {\n \"role\":
\"assistant\",\n \"content\": \"Test Task: \\n\\nWrite a program in
your preferred programming language that accomplishes the following:\\n\\n1.
Accepts a list of integers as input.\\n2. Filters the list to include only
prime numbers.\\n3. Sorts the filtered prime numbers in ascending order.\\n4.
Outputs the sorted list of prime numbers.\\n\\nRequirements:\\n- Provide the
complete source code.\\n- Include comments explaining the logic.\\n- Ensure
the code handles invalid inputs gracefully.\\n- Test the program with at least
three different input sets and show the outputs.\\n\\nExample:\\n\\nInput:
[12, 7, 5, 18, 11, 3, 20]\\nOutput: [3, 5, 7, 11]\\n\\nPlease provide the
full code along with sample input and output for verification.\",\n \"refusal\":
null,\n \"annotations\": []\n },\n \"logprobs\": null,\n
\ \"finish_reason\": \"stop\"\n }\n ],\n \"usage\": {\n \"prompt_tokens\":
147,\n \"completion_tokens\": 157,\n \"total_tokens\": 304,\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_e76a310957\"\n}\n"
string: "{\n \"id\": \"chatcmpl-AmjYyKbTn42DzaLVOjDvJpLubTjSq\",\n \"object\"\
: \"chat.completion\",\n \"created\": 1736178252,\n \"model\": \"gpt-4o-2024-08-06\"\
,\n \"choices\": [\n {\n \"index\": 0,\n \"message\": {\n \
\ \"role\": \"assistant\",\n \"content\": \"Action: Another Test\
\ Tool\\nAction Input: {\\\"query\\\": \\\"AI and AI agents\\\"}\",\n \
\ \"refusal\": null\n },\n \"logprobs\": null,\n \"finish_reason\"\
: \"stop\"\n }\n ],\n \"usage\": {\n \"prompt_tokens\": 295,\n \
\ \"completion_tokens\": 18,\n \"total_tokens\": 313,\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 \"system_fingerprint\": \"fp_5f20662549\"\n}\n"
headers:
CF-Cache-Status:
- DYNAMIC
CF-Ray:
- 9dc819faed93433f-EWR
CF-RAY:
- 8fdcd3fc9a56bf66-ATL
Connection:
- keep-alive
Content-Type:
- application/json
Date:
- Sun, 15 Mar 2026 02:31:26 GMT
- Mon, 06 Jan 2025 15:44:12 GMT
Server:
- cloudflare
Strict-Transport-Security:
- STS-XXX
Set-Cookie:
- __cf_bm=X1fuDKrQrN8tU.uxjB0murgJXWXcPtlNLnD7xUrAKTs-1736178252-1.0.1.1-AME9VZZVtEpqX9.BEN_Kj9pI9uK3sIJc2LdbuPsP3wULKxF4Il6r8ghX0to2wpcYsGWbJXSqWP.dQz4vGf_Gbw;
path=/; expires=Mon, 06-Jan-25 16:14:12 GMT; domain=.api.openai.com; HttpOnly;
Secure; SameSite=None
- _cfuvid=mv42xOepGYaNopc5ovT9Ajamw5rJrze8tlWTik8lfrk-1736178252935-0.0.1.1-604800000;
path=/; domain=.api.openai.com; HttpOnly; Secure; SameSite=None
Transfer-Encoding:
- chunked
X-Content-Type-Options:
- X-CONTENT-TYPE-XXX
- nosniff
access-control-expose-headers:
- ACCESS-CONTROL-XXX
- X-Request-ID
alt-svc:
- h3=":443"; ma=86400
openai-organization:
- OPENAI-ORG-XXX
- crewai-iuxna1
openai-processing-ms:
- '3681'
openai-project:
- OPENAI-PROJECT-XXX
- '632'
openai-version:
- '2020-10-01'
set-cookie:
- SET-COOKIE-XXX
x-openai-proxy-wasm:
- v0.1
strict-transport-security:
- max-age=31536000; includeSubDomains; preload
x-ratelimit-limit-requests:
- X-RATELIMIT-LIMIT-REQUESTS-XXX
- '10000'
x-ratelimit-limit-tokens:
- X-RATELIMIT-LIMIT-TOKENS-XXX
- '30000000'
x-ratelimit-remaining-requests:
- X-RATELIMIT-REMAINING-REQUESTS-XXX
- '9999'
x-ratelimit-remaining-tokens:
- X-RATELIMIT-REMAINING-TOKENS-XXX
- '29999644'
x-ratelimit-reset-requests:
- X-RATELIMIT-RESET-REQUESTS-XXX
- 6ms
x-ratelimit-reset-tokens:
- X-RATELIMIT-RESET-TOKENS-XXX
- 0s
x-request-id:
- X-REQUEST-ID-XXX
- req_9276753b2200fc95c74fc43c9d7d84a6
status:
code: 200
message: OK
- request:
body: '{"messages": [{"role": "system", "content": "You are Researcher. You''re
an expert researcher, specialized in technology, software engineering, AI and
startups. You work as a freelancer and is now working on doing research and
analysis for a new customer.\nYour personal goal is: Make the best research
and analysis on content about AI and AI agents\nYou ONLY have access to the
following tools, and should NEVER make up tools that are not listed here:\n\nTool
Name: Another Test Tool\nTool Arguments: {''query'': {''description'': ''Query
to process'', ''type'': ''str''}}\nTool Description: Another test tool\n\nUse
the following format:\n\nThought: you should always think about what to do\nAction:
the action to take, only one name of [Another Test Tool], just the name, exactly
as it''s written.\nAction Input: the input to the action, just a simple python
dictionary, enclosed in curly braces, using \" to wrap keys and values.\nObservation:
the result of the action\n\nOnce all necessary information is gathered:\n\nThought:
I now know the final answer\nFinal Answer: the final answer to the original
input question"}, {"role": "user", "content": "\nCurrent Task: Write a test
task\n\nThis is the expect criteria for your final answer: Test output\nyou
MUST return the actual complete content as the final answer, not a summary.\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:"}, {"role": "assistant", "content":
"Action: Another Test Tool\nAction Input: {\"query\": \"AI and AI agents\"}\nObservation:
Another processed: AI and AI agents"}], "model": "gpt-4o", "stop": ["\nObservation:"],
"stream": false}'
headers:
accept:
- application/json
accept-encoding:
- gzip, deflate
connection:
- keep-alive
content-length:
- '1687'
content-type:
- application/json
cookie:
- _cfuvid=mv42xOepGYaNopc5ovT9Ajamw5rJrze8tlWTik8lfrk-1736178252935-0.0.1.1-604800000;
__cf_bm=X1fuDKrQrN8tU.uxjB0murgJXWXcPtlNLnD7xUrAKTs-1736178252-1.0.1.1-AME9VZZVtEpqX9.BEN_Kj9pI9uK3sIJc2LdbuPsP3wULKxF4Il6r8ghX0to2wpcYsGWbJXSqWP.dQz4vGf_Gbw
host:
- api.openai.com
user-agent:
- OpenAI/Python 1.52.1
x-stainless-arch:
- arm64
x-stainless-async:
- 'false'
x-stainless-lang:
- python
x-stainless-os:
- MacOS
x-stainless-package-version:
- 1.52.1
x-stainless-raw-response:
- 'true'
x-stainless-retry-count:
- '0'
x-stainless-runtime:
- CPython
x-stainless-runtime-version:
- 3.12.7
method: POST
uri: https://api.openai.com/v1/chat/completions
response:
body:
string: "{\n \"id\": \"chatcmpl-AmjYzChV9s4D4qOJJvTvBAt3kRh7n\",\n \"object\"\
: \"chat.completion\",\n \"created\": 1736178253,\n \"model\": \"gpt-4o-2024-08-06\"\
,\n \"choices\": [\n {\n \"index\": 0,\n \"message\": {\n \
\ \"role\": \"assistant\",\n \"content\": \"Thought: I now know\
\ the final answer\\nFinal Answer: Another processed: AI and AI agents\",\n\
\ \"refusal\": null\n },\n \"logprobs\": null,\n \"\
finish_reason\": \"stop\"\n }\n ],\n \"usage\": {\n \"prompt_tokens\"\
: 326,\n \"completion_tokens\": 19,\n \"total_tokens\": 345,\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 \"system_fingerprint\"\
: \"fp_5f20662549\"\n}\n"
headers:
CF-Cache-Status:
- DYNAMIC
CF-RAY:
- 8fdcd4011938bf66-ATL
Connection:
- keep-alive
Content-Type:
- application/json
Date:
- Mon, 06 Jan 2025 15:44:15 GMT
Server:
- cloudflare
Transfer-Encoding:
- chunked
X-Content-Type-Options:
- nosniff
access-control-expose-headers:
- X-Request-ID
alt-svc:
- h3=":443"; ma=86400
openai-organization:
- crewai-iuxna1
openai-processing-ms:
- '2488'
openai-version:
- '2020-10-01'
strict-transport-security:
- max-age=31536000; includeSubDomains; preload
x-ratelimit-limit-requests:
- '10000'
x-ratelimit-limit-tokens:
- '30000000'
x-ratelimit-remaining-requests:
- '9999'
x-ratelimit-remaining-tokens:
- '29999613'
x-ratelimit-reset-requests:
- 6ms
x-ratelimit-reset-tokens:
- 0s
x-request-id:
- req_5e3a1a90ef91ff4f12d5b84e396beccc
status:
code: 200
message: OK

File diff suppressed because one or more lines are too long