langgraph-checkpoint-mongodb

What it is and what it does

This package implements a CheckpointSaver for LangGraph that stores agent state and execution history in MongoDB. It is designed for workflows that need to persist intermediate steps, recover from failures, or replay execution paths—common requirements in multi-step agentic applications.

The package provides both synchronous and asynchronous APIs (put/get/list and aput/aget/alist) for storing, retrieving, and listing checkpoints. It integrates with LangGraph's checkpoint protocol and relies on pymongo for MongoDB connectivity and langchain-mongodb for higher-level abstractions. You instantiate a MongoDBSaver from a connection string, then use it as a context manager to persist and recover checkpoint data.

Use it for:

• Store intermediate states in multi-step workflows to enable resumption after failures.
• Persist agent decision history and channel values for audit trails and debugging.
• Implement checkpoint-based recovery in long-running tasks without restarting from scratch.
• Build async-first systems that checkpoint state asynchronously during execution.
• Retrieve and analyze past execution paths using MongoDB queries.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Provides a MongoDB-backed checkpoint storage implementation for LangGraph, enabling state persistence and recovery in agentic workflows.

Yes, if you are building LangGraph applications that require MongoDB-backed state persistence. Install friction is low, maintenance is active, and the API is straightforward. Verify the license terms first and confirm MongoDB availability in your deployment environment.

Install

langgraph-checkpoint-mongodb on PyPI

pip install langgraph-checkpoint-mongodb

uv add langgraph-checkpoint-mongodb

poetry add langgraph-checkpoint-mongodb

Installing langgraph-checkpoint-mongodb

Before you install

Low friction installation with three runtime dependencies. Package is actively maintained and requires Python 3.10 or later.

License in practice

License status is unclear; the package metadata does not declare an SPDX identifier or license text. Verify the actual license terms before use.

Quickstart

pip install langgraph-checkpoint-mongodb

from langgraph.checkpoint.mongodb import MongoDBSaver

MONGODB_URI = "mongodb://localhost:27017"
DB_NAME = "checkpoint_example"

with MongoDBSaver.from_conn_string(MONGODB_URI, DB_NAME) as checkpointer:
    checkpointer.put(write_config, checkpoint, {}, {})
    checkpointer.get(read_config)

Requires a running MongoDB instance and Python 3.10 or later.

Verify before relying

• Exact license terms and any restrictions on commercial or proprietary use.
• Production-readiness and stability guarantees for the package.

Package facts

Evidence: langgraph_checkpoint_mongodb-0.4.0-py3-none-any.whl

Tags

More Database packages