Feat/memory base (#1444)

* byom - short/entity memory

* better

* rm uneeded

* fix text

* use context

* rm dep and sync

* type check fix

* fixed test using new cassete

* fixing types

* fixed types

* fix types

* fixed types

* fixing types

* fix type

* cassette update

* just mock the return of short term mem

* remove print

* try catch block

* added docs

* dding error handling here

src/crewai/agents/agent_builder/base_agent_executor_mixin.py

@@ -17,7 +17,7 @@ if TYPE_CHECKING:
 
 class CrewAgentExecutorMixin:
     crew: Optional["Crew"]
-    crew_agent: Optional["BaseAgent"]
+    agent: Optional["BaseAgent"]
     task: Optional["Task"]
     iterations: int
     have_forced_answer: bool
@@ -33,9 +33,9 @@ class CrewAgentExecutorMixin:
         """Create and save a short-term memory item if conditions are met."""
         if (
             self.crew
-            and self.crew_agent
+            and self.agent
             and self.task
-            and "Action: Delegate work to coworker" not in output.log
+            and "Action: Delegate work to coworker" not in output.text
         ):
             try:
                 if (
@@ -43,11 +43,11 @@ class CrewAgentExecutorMixin:
                     and self.crew._short_term_memory
                 ):
                     self.crew._short_term_memory.save(
-                        value=output.log,
+                        value=output.text,
                         metadata={
                             "observation": self.task.description,
                         },
-                        agent=self.crew_agent.role,
+                        agent=self.agent.role,
                     )
             except Exception as e:
                 print(f"Failed to add to short term memory: {e}")
@@ -61,18 +61,18 @@ class CrewAgentExecutorMixin:
             and self.crew._long_term_memory
             and self.crew._entity_memory
             and self.task
-            and self.crew_agent
+            and self.agent
         ):
             try:
-                ltm_agent = TaskEvaluator(self.crew_agent)
-                evaluation = ltm_agent.evaluate(self.task, output.log)
+                ltm_agent = TaskEvaluator(self.agent)
+                evaluation = ltm_agent.evaluate(self.task, output.text)
 
                 if isinstance(evaluation, ConverterError):
                     return
 
                 long_term_memory = LongTermMemoryItem(
                     task=self.task.description,
-                    agent=self.crew_agent.role,
+                    agent=self.agent.role,
                     quality=evaluation.quality,
                     datetime=str(time.time()),
                     expected_output=self.task.expected_output,

src/crewai/agents/crew_agent_executor.py

@@ -19,6 +19,7 @@ from crewai.utilities.exceptions.context_window_exceeding_exception import (
 )
 from crewai.utilities.logger import Logger
 from crewai.utilities.training_handler import CrewTrainingHandler
+from crewai.agents.agent_builder.base_agent import BaseAgent
 
 
 class CrewAgentExecutor(CrewAgentExecutorMixin):
@@ -29,7 +30,7 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
         llm: Any,
         task: Any,
         crew: Any,
-        agent: Any,
+        agent: BaseAgent,
         prompt: dict[str, str],
         max_iter: int,
         tools: List[Any],
@@ -103,7 +104,8 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
 
             if self.crew and self.crew._train:
                 self._handle_crew_training_output(formatted_answer)
-
+        self._create_short_term_memory(formatted_answer)
+        self._create_long_term_memory(formatted_answer)
         return {"output": formatted_answer.output}
 
     def _invoke_loop(self, formatted_answer=None):
@@ -176,6 +178,8 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
         return formatted_answer
 
     def _show_start_logs(self):
+        if self.agent is None:
+            raise ValueError("Agent cannot be None")
         if self.agent.verbose or (
             hasattr(self, "crew") and getattr(self.crew, "verbose", False)
         ):
@@ -188,6 +192,8 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
             )
 
     def _show_logs(self, formatted_answer: Union[AgentAction, AgentFinish]):
+        if self.agent is None:
+            raise ValueError("Agent cannot be None")
         if self.agent.verbose or (
             hasattr(self, "crew") and getattr(self.crew, "verbose", False)
         ):
@@ -306,7 +312,7 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
         self, result: AgentFinish, human_feedback: str | None = None
     ) -> None:
         """Function to handle the process of the training data."""
-        agent_id = str(self.agent.id)
+        agent_id = str(self.agent.id)  # type: ignore
 
         # Load training data
         training_handler = CrewTrainingHandler(TRAINING_DATA_FILE)
@@ -339,7 +345,7 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                 "initial_output": result.output,
                 "human_feedback": human_feedback,
                 "agent": agent_id,
-                "agent_role": self.agent.role,
+                "agent_role": self.agent.role,  # type: ignore
             }
             if self.crew is not None and hasattr(self.crew, "_train_iteration"):
                 train_iteration = self.crew._train_iteration

src/crewai/crew.py

@@ -126,8 +126,8 @@ class Crew(BaseModel):
         default=None,
         description="An Instance of the EntityMemory to be used by the Crew",
     )
-    embedder: Optional[dict] = Field(
-        default={"provider": "openai"},
+    embedder: Optional[Any] = Field(
+        default=None,
         description="Configuration for the embedder to be used for the crew.",
     )
     usage_metrics: Optional[UsageMetrics] = Field(
@@ -774,7 +774,9 @@ class Crew(BaseModel):
 
     def _log_task_start(self, task: Task, role: str = "None"):
         if self.output_log_file:
-            self._file_handler.log(task_name=task.name, task=task.description, agent=role, status="started")
+            self._file_handler.log(
+                task_name=task.name, task=task.description, agent=role, status="started"
+            )
 
     def _update_manager_tools(self, task: Task):
         if self.manager_agent:
@@ -796,7 +798,13 @@ class Crew(BaseModel):
     def _process_task_result(self, task: Task, output: TaskOutput) -> None:
         role = task.agent.role if task.agent is not None else "None"
         if self.output_log_file:
-            self._file_handler.log(task_name=task.name, task=task.description, agent=role, status="completed", output=output.raw)
+            self._file_handler.log(
+                task_name=task.name,
+                task=task.description,
+                agent=role,
+                status="completed",
+                output=output.raw,
+            )
 
     def _create_crew_output(self, task_outputs: List[TaskOutput]) -> CrewOutput:
         if len(task_outputs) != 1:

src/crewai/memory/contextual/contextual_memory.py

@@ -31,7 +31,9 @@ class ContextualMemory:
         formatted as bullet points.
         """
         stm_results = self.stm.search(query)
-        formatted_results = "\n".join([f"- {result}" for result in stm_results])
+        formatted_results = "\n".join(
+            [f"- {result['context']}" for result in stm_results]
+        )
         return f"Recent Insights:\n{formatted_results}" if stm_results else ""
 
     def _fetch_ltm_context(self, task) -> Optional[str]:

src/crewai/memory/long_term/long_term_memory.py

@@ -1,4 +1,4 @@
-from typing import Any, Dict
+from typing import Any, Dict, List
 
 from crewai.memory.long_term.long_term_memory_item import LongTermMemoryItem
 from crewai.memory.memory import Memory
@@ -28,7 +28,7 @@ class LongTermMemory(Memory):
             datetime=item.datetime,
         )
 
-    def search(self, task: str, latest_n: int = 3) -> Dict[str, Any]:
+    def search(self, task: str, latest_n: int = 3) -> List[Dict[str, Any]]:  # type: ignore # signature of "search" incompatible with supertype "Memory"
         return self.storage.load(task, latest_n)  # type: ignore # BUG?: "Storage" has no attribute "load"
 
     def reset(self) -> None:

src/crewai/memory/memory.py

@@ -1,6 +1,6 @@
-from typing import Any, Dict, Optional
+from typing import Any, Dict, Optional, List
 
-from crewai.memory.storage.interface import Storage
+from crewai.memory.storage.rag_storage import RAGStorage
 
 
 class Memory:
@@ -8,7 +8,7 @@ class Memory:
     Base class for memory, now supporting agent tags and generic metadata.
     """
 
-    def __init__(self, storage: Storage):
+    def __init__(self, storage: RAGStorage):
         self.storage = storage
 
     def save(
@@ -23,5 +23,5 @@ class Memory:
 
         self.storage.save(value, metadata)
 
-    def search(self, query: str) -> Dict[str, Any]:
+    def search(self, query: str) -> List[Dict[str, Any]]:
         return self.storage.search(query)

src/crewai/memory/storage/base_rag_storage.py

@@ -0,0 +1,76 @@
+from abc import ABC, abstractmethod
+from typing import Any, Dict, List, Optional
+
+
+class BaseRAGStorage(ABC):
+    """
+    Base class for RAG-based Storage implementations.
+    """
+
+    app: Any | None = None
+
+    def __init__(
+        self,
+        type: str,
+        allow_reset: bool = True,
+        embedder_config: Optional[Any] = None,
+        crew: Any = None,
+    ):
+        self.type = type
+        self.allow_reset = allow_reset
+        self.embedder_config = embedder_config
+        self.crew = crew
+        self.agents = self._initialize_agents()
+
+    def _initialize_agents(self) -> str:
+        if self.crew:
+            return "_".join(
+                [self._sanitize_role(agent.role) for agent in self.crew.agents]
+            )
+        return ""
+
+    @abstractmethod
+    def _sanitize_role(self, role: str) -> str:
+        """Sanitizes agent roles to ensure valid directory names."""
+        pass
+
+    @abstractmethod
+    def save(self, value: Any, metadata: Dict[str, Any]) -> None:
+        """Save a value with metadata to the storage."""
+        pass
+
+    @abstractmethod
+    def search(
+        self,
+        query: str,
+        limit: int = 3,
+        filter: Optional[dict] = None,
+        score_threshold: float = 0.35,
+    ) -> List[Any]:
+        """Search for entries in the storage."""
+        pass
+
+    @abstractmethod
+    def reset(self) -> None:
+        """Reset the storage."""
+        pass
+
+    @abstractmethod
+    def _generate_embedding(
+        self, text: str, metadata: Optional[Dict[str, Any]] = None
+    ) -> Any:
+        """Generate an embedding for the given text and metadata."""
+        pass
+
+    @abstractmethod
+    def _initialize_app(self):
+        """Initialize the vector db."""
+        pass
+
+    def setup_config(self, config: Dict[str, Any]):
+        """Setup the config of the storage."""
+        pass
+
+    def initialize_client(self):
+        """Initialize the client of the storage. This should setup the app and the db collection"""
+        pass

src/crewai/memory/storage/interface.py

@@ -1,4 +1,4 @@
-from typing import Any, Dict
+from typing import Any, Dict, List
 
 
 class Storage:
@@ -7,7 +7,7 @@ class Storage:
     def save(self, value: Any, metadata: Dict[str, Any]) -> None:
         pass
 
-    def search(self, key: str) -> Dict[str, Any]:  # type: ignore
+    def search(self, key: str) -> List[Dict[str, Any]]:  # type: ignore
         pass
 
     def reset(self) -> None:

src/crewai/memory/storage/rag_storage.py

@@ -3,10 +3,11 @@ import io
 import logging
 import os
 import shutil
+import uuid
 from typing import Any, Dict, List, Optional
-
-from crewai.memory.storage.interface import Storage
+from crewai.memory.storage.base_rag_storage import BaseRAGStorage
 from crewai.utilities.paths import db_storage_path
+from chromadb.api import ClientAPI
 
 
 @contextlib.contextmanager
@@ -24,61 +25,42 @@ def suppress_logging(
     logger.setLevel(original_level)
 
 
-class RAGStorage(Storage):
+class RAGStorage(BaseRAGStorage):
     """
     Extends Storage to handle embeddings for memory entries, improving
     search efficiency.
     """
 
-    def __init__(self, type, allow_reset=True, embedder_config=None, crew=None):
-        super().__init__()
-        if (
-            not os.getenv("OPENAI_API_KEY")
-            and not os.getenv("OPENAI_BASE_URL") == "https://api.openai.com/v1"
-        ):
-            os.environ["OPENAI_API_KEY"] = "fake"
+    app: ClientAPI | None = None
 
+    def __init__(self, type, allow_reset=True, embedder_config=None, crew=None):
+        super().__init__(type, allow_reset, embedder_config, crew)
         agents = crew.agents if crew else []
         agents = [self._sanitize_role(agent.role) for agent in agents]
         agents = "_".join(agents)
+        self.agents = agents
 
-        config = {
-            "app": {
-                "config": {"name": type, "collect_metrics": False, "log_level": "ERROR"}
-            },
-            "chunker": {
-                "chunk_size": 5000,
-                "chunk_overlap": 100,
-                "length_function": "len",
-                "min_chunk_size": 150,
-            },
-            "vectordb": {
-                "provider": "chroma",
-                "config": {
-                    "collection_name": type,
-                    "dir": f"{db_storage_path()}/{type}/{agents}",
-                    "allow_reset": allow_reset,
-                },
-            },
-        }
-
-        if embedder_config:
-            config["embedder"] = embedder_config
         self.type = type
-        self.config = config
+        self.embedder_config = embedder_config or self._create_embedding_function()
         self.allow_reset = allow_reset
+        self._initialize_app()
 
     def _initialize_app(self):
-        from embedchain import App
-        from embedchain.llm.base import BaseLlm
+        import chromadb
 
-        class FakeLLM(BaseLlm):
-            pass
+        chroma_client = chromadb.PersistentClient(
+            path=f"{db_storage_path()}/{self.type}/{self.agents}"
+        )
+        self.app = chroma_client
 
-        self.app = App.from_config(config=self.config)
-        self.app.llm = FakeLLM()
-        if self.allow_reset:
-            self.app.reset()
+        try:
+            self.collection = self.app.get_collection(
+                name=self.type, embedding_function=self.embedder_config
+            )
+        except Exception:
+            self.collection = self.app.create_collection(
+                name=self.type, embedding_function=self.embedder_config
+            )
 
     def _sanitize_role(self, role: str) -> str:
         """
@@ -87,11 +69,14 @@ class RAGStorage(Storage):
         return role.replace("\n", "").replace(" ", "_").replace("/", "_")
 
     def save(self, value: Any, metadata: Dict[str, Any]) -> None:
-        if not hasattr(self, "app"):
+        if not hasattr(self, "app") or not hasattr(self, "collection"):
             self._initialize_app()
-        self._generate_embedding(value, metadata)
+        try:
+            self._generate_embedding(value, metadata)
+        except Exception as e:
+            logging.error(f"Error during {self.type} save: {str(e)}")
 
-    def search(  # type: ignore # BUG?: Signature of "search" incompatible with supertype "Storage"
+    def search(
         self,
         query: str,
         limit: int = 3,
@@ -100,31 +85,50 @@ class RAGStorage(Storage):
     ) -> List[Any]:
         if not hasattr(self, "app"):
             self._initialize_app()
-        from embedchain.vectordb.chroma import InvalidDimensionException
 
-        with suppress_logging():
-            try:
-                results = (
-                    self.app.search(query, limit, where=filter)
-                    if filter
-                    else self.app.search(query, limit)
-                )
-            except InvalidDimensionException:
-                self.app.reset()
-                return []
-        return [r for r in results if r["metadata"]["score"] >= score_threshold]
+        try:
+            with suppress_logging():
+                response = self.collection.query(query_texts=query, n_results=limit)
 
-    def _generate_embedding(self, text: str, metadata: Dict[str, Any]) -> Any:
-        if not hasattr(self, "app"):
+            results = []
+            for i in range(len(response["ids"][0])):
+                result = {
+                    "id": response["ids"][0][i],
+                    "metadata": response["metadatas"][0][i],
+                    "context": response["documents"][0][i],
+                    "score": response["distances"][0][i],
+                }
+                if result["score"] >= score_threshold:
+                    results.append(result)
+
+            return results
+        except Exception as e:
+            logging.error(f"Error during {self.type} search: {str(e)}")
+            return []
+
+    def _generate_embedding(self, text: str, metadata: Dict[str, Any]) -> None:  # type: ignore
+        if not hasattr(self, "app") or not hasattr(self, "collection"):
             self._initialize_app()
-        from embedchain.models.data_type import DataType
 
-        self.app.add(text, data_type=DataType.TEXT, metadata=metadata)
+        self.collection.add(
+            documents=[text],
+            metadatas=[metadata or {}],
+            ids=[str(uuid.uuid4())],
+        )
 
     def reset(self) -> None:
         try:
             shutil.rmtree(f"{db_storage_path()}/{self.type}")
+            if self.app:
+                self.app.reset()
         except Exception as e:
             raise Exception(
                 f"An error occurred while resetting the {self.type} memory: {e}"
             )
+
+    def _create_embedding_function(self):
+        import chromadb.utils.embedding_functions as embedding_functions
+
+        return embedding_functions.OpenAIEmbeddingFunction(
+            api_key=os.getenv("OPENAI_API_KEY"), model_name="text-embedding-3-small"
+        )