Skip to main content

memory.base

Base class for memory backends.

BaseMemoryBackend Objects​

class BaseMemoryBackend(ABC)

Abstract memory backend.

Memory backends provide persistent context for agents across conversations. They support adding memories, searching them, and auto-capturing conversation context.

add​

@abstractmethod
async def add(messages: list[dict],
              metadata: dict[str, Any] | None = None) -> None

Add messages to memory.

Parameters​

messages : list[dict] Conversation messages ({"role": ..., "content": ...}). metadata : dict | None Optional metadata (topic, tools used, outcome, etc.).

@abstractmethod
async def search(query: str, limit: int = 10) -> list[dict[str, Any]]

Search memory for relevant entries.

Parameters​

query : str Search query text. limit : int Maximum number of results to return.

Returns​

list[dict] Matching memory entries with content and score keys.

get_relevant_context​

async def get_relevant_context(query: str, max_tokens: int = 2000) -> str

Get relevant memory context for system prompt injection.

Parameters​

query : str The current user message or task description. max_tokens : int Approximate maximum tokens for the returned context.

Returns​

str Formatted memory context for injection into the system prompt.

close​

async def close() -> None

Close the memory backend and release resources.