Memory API Reference¶

ConversationMemory¶

In-memory conversation message storage with optional sliding window.

from flowgentra_ai import ConversationMemory

ConversationMemory(max_messages: int | None = None)

Method	Returns	Description
`add_message(thread_id, message)`	`None`	Add a message to a thread
`messages(thread_id, limit=None)`	`list[Message]`	Get messages (oldest first)
`clear(thread_id)`	`None`	Clear all messages for a thread

A saved state checkpoint for persistence/resume.

from flowgentra_ai import Checkpoint

Property	Type	Description
`metadata`	`CheckpointMetadata`	Checkpoint metadata
`state`	`dict`	Saved state data

Metadata for a checkpoint.

Property	Type	Description
`last_node`	`str \\| None`	Last executed node
`execution_path`	`list[str]`	Ordered list of executed nodes
`extra`	`dict`	Additional metadata

File-based checkpoint storage.

from flowgentra_ai import FileCheckpointer

Typically used via StateGraphBuilder.set_checkpointer(base_dir) rather than directly.

Conversation memory that stays within a token budget.

from flowgentra_ai import TokenBufferMemory

Memory that automatically summarizes older messages.

from flowgentra_ai import SummaryMemory, SummaryConfig