memory

Agent Memory Plugin

Attributes

Classes

MemoryPlugin

Bases: BaseAIPlugin

Agent memory capabilities plugin.

Functions

after_new_context_created

after_new_context_created(
    request: ChatCompletionRequest,
    sa_session: SASession,
    neo4j_session: NJSession,
    context: EleanorAIRequestContext,
) -> None

Triggered when a new AI request context is created

Parameters:

• request (ChatCompletionRequest) –

  The chat completion request.

• session (SASession) –

  The SQLAlchemy session object.

• context (EleanorAIRequestContext) –

  The request context.

Returns:

• None –

  None

after_response_generation

after_response_generation(
    request: ChatCompletionRequest,
    sa_session: SASession,
    neo4j_session: NJSession,
    context: EleanorAIRequestContext,
    response_txt_n: List[str],
) -> None

Executes after all LLM responses have been generated

Parameters:

• request (ChatCompletionRequest) –

  The chat completion request.

• session (SASession) –

  The SQLAlchemy session object.

• context (EleanorAIRequestContext) –

  The AI request context.

• response_txt_n (List[str]) –

  The list of generated response texts.

Returns:

• None –

  None

before_prompt_rendered

before_prompt_rendered(
    request: ChatCompletionRequest,
    sa_session: SASession,
    neo4j_session: NJSession,
    context: EleanorAIRequestContext,
    chat: CanonicalChat,
) -> CanonicalChat

Retrieves agent memories and injects them into chat object

Triggered before the CanonicalChat object is rendered into the LLM-specific prompt string.

Parameters:

• request (ChatCompletionRequest) –

  The completion request object.

• session (SASession) –

  The SQLAlchemy session object.

• context (EleanorAIRequestContext) –

  The AI request context.

• chat (CanonicalChat) –

  The chat object.

Returns:

• CanonicalChat ( CanonicalChat ) –

  The updated chat object.

conversation_insights

conversation_insights(
    sa_session: SASession,
    neo4j_session: NJSession,
    namespace_pkid: str,
    agent_pkid: str,
    conversation: str,
    insight_count: int,
    existing_memories: str,
    context: EleanorAIRequestContext,
) -> List[str]

Retrieves conversation insights for a given agent and conversation.

Parameters:

• agent_pkid (str) –

  The primary key of the agent.

• conversation (str) –

  The conversation text.

• insight_count (int) –

  The number of insights to retrieve.

Returns:

• List[str] –

  List[str]: A list of conversation insights.

Raises:

• ServiceError –

  If no insight responses are returned from the conversation insights chain or if there is an error invoking the conversation insights chain.

Functions