LangChain Releases OpenWiki: Open-Source Tool That Automatically Documents Repositories for AI Agents

LangChain has released OpenWiki — an open-source CLI tool that automatically generates and maintains documentation wikis for code repositories so that AI agents have structured context about the project they are working on. The repository is available at github.com/langchain-ai/openwiki.

Why Do Agents Need Structured Repository Documentation?

One of the chronic problems with modern coding agents is that they need context about the repository — which modules it contains, how they are organized, what the key conventions are — but delivering that context is not trivial. The common approach is to manually write and update instruction files like AGENTS.md or CLAUDE.md. That approach has two problems: manual maintenance falls behind code changes, and bloated instruction files waste the agent’s precious context space unnecessarily.

OpenWiki solves this through structure: documentation is generated as a separate wiki, and instruction files contain only referenced pointers to it. The agent fetches wiki content on demand, as needed, instead of loading everything upfront.

How the Tool Works

OpenWiki is built on LangChain’s DeepAgents framework and integrates LangSmith tracing for inspecting the agent’s work during documentation generation and updates — useful for diagnosing and optimizing the process.

Initial wiki generation is triggered with two commands:

npm install -g openwiki
openwiki --init

The agent analyzes the repository structure and generates documentation organized as a structured wiki, then updates AGENTS.md or CLAUDE.md with references to the new resource. From that point on, agents have access to relevant context without the entire documentation living inline in the instruction file.

Updates That Follow the Live Codebase

The key advantage of OpenWiki over one-shot documentation generation tools is its update mechanism. The tool ships with a GitHub Action that runs on a schedule and analyzes recent commits and git diffs. Based on that analysis it incrementally refreshes only the parts of the documentation that have changed — without fully regenerating the entire wiki. This means lower LLM call costs and faster updates for active repositories with frequent commits.

Support for Multiple LLM Providers

OpenWiki supports OpenRouter as the default provider (with open models), plus Fireworks, Baseten, OpenAI, and Anthropic. This flexibility allows teams to use the provider they already have or that fits the project’s cost model — without needing to adapt the tool’s code.

The optional LangSmith integration gives developers full visibility into what the agent reads and reasons about while generating or updating documentation, which is especially useful when tuning the quality of the output wiki pages.

Who Is OpenWiki For?

The primary users are development teams already using AI coding agents — especially those who have noticed that agents make mistakes due to outdated or missing repository documentation. The tool is complementary to existing tools: it does not replace AGENTS.md or CLAUDE.md, but offloads their verbose content into a separate, automatically maintained wiki.

For teams with multiple repositories or monorepos, the scheduled GitHub Actions runs offer a clear path toward automated, always-current documentation infrastructure with no manual overhead.