🦜️🔗 LangChain Cloudflare

This is a Monorepo containing LangChain packages for Cloudflare. It includes integrations between WorkersAI, AI Search, Vectorize, D1, LangChain, and LangGraph.

It contains the following packages.

• langchain-cloudflare (PyPI)
• langgraph-checkpoint-cloudflare-d1 (PyPI)
• langmem-cloudflare-vectorize (PyPI)

Features

LangChain

Components

• ChatCloudflareWorkersAI
• CloudflareWorkersAIEmbeddings
• CloudflareWorkersAIReranker — document reranking on Workers AI
• CloudflareVectorize
• CloudflareAISearchRetriever — Cloudflare AI Search (fka AutoRAG)
• CloudflareBrowserRunLoader — document loader via Browser Run
• CloudflareBrowserRunTool — agent tool via Browser Run

LangGraph

• Checkpointing (BaseCheckpointSaver)
  • CloudflareD1Saver
  • AsyncCloudflareD1Saver
• MemoryStore (BaseStore)
  • CloudflareVectorizeBaseStore

Installation

You can install the langchain-cloudflare package from PyPI.

pip install langchain-cloudflare

You can install the langgraph-checkpoint-cloudflare-d1 package from PyPI as well:

pip install langgraph-checkpoint-cloudflare-d1

And to install the langmem-cloudflare-vectorize package from PyPI:

pip install langmem-cloudflare-vectorize

Usage

See langchain-cloudflare usage , langgraph-checkpoint-cloudflare-d1 usage. and langmem-cloudflare-vectorize usage

Example notebooks

Example notebooks use a dedicated, locked environment with this checkout of langchain-cloudflare installed. From the repository root:

make notebook_sync
make notebook_check
make notebook_lab

notebook_lab loads credentials from the repo-root .env before starting JupyterLab. Open a notebook under docs/, such as ai_search.ipynb or browser_run.ipynb, and run it manually. Stop the JupyterLab process with Ctrl-C.

License

This project is licensed under the MIT License.