OpenWiki is a command-line interface tool developed by LangChain AI that automates the creation and maintenance of documentation for software codebases, with a specific focus on supporting AI agents. Written in TypeScript, the tool addresses the challenge of keeping documentation synchronized with evolving codebases by leveraging language models to generate and update documentation automatically.
The core functionality of OpenWiki centers on documentation generation and maintenance workflows. When initialized in a repository, the tool creates an openwiki directory containing generated documentation. For repositories that already have documentation in place, OpenWiki refreshes and updates that documentation based on detected repository changes. The CLI operates in an interactive mode by default, allowing developers to send follow-up messages and refine documentation through conversation. For automated workflows, the tool supports a non-interactive print mode using the -p or --print flags, which executes a single command and exits after outputting results.
A distinctive feature of OpenWiki is its integration with agent-based development workflows. The tool automatically appends prompting instructions to AGENTS.md and CLAUDE.md files within repositories, directing coding agents to reference the generated documentation when searching for context. If these files do not exist, OpenWiki creates them automatically. This design choice reflects the tool's purpose of serving as a bridge between human-written code and AI agents that need to understand and work with that code.
OpenWiki supports multiple inference providers and language models out of the box, including OpenRouter, Fireworks, Baseten, OpenAI, Anthropic, and OpenAI-compatible endpoints. The tool comes with several pre-configured models such as GLM 5.2, Kimi K2.6, and Sonnet 5, while also allowing users to specify custom model IDs for each provider. This flexibility enables teams to choose their preferred LLM infrastructure. The tool supports alternative base URLs for Anthropic-compatible endpoints, allowing users to route requests through self-hosted or proxied gateways. For OpenAI-compatible endpoints, users can configure any OpenAI-shaped API by specifying a required base URL.
Initial setup involves an interactive configuration process where users specify their inference provider, API key, and preferred language model. Optionally, users can configure a LangSmith API key to trace OpenWiki runs to a LangSmith project named openwiki. All configuration options and secrets are stored locally in ~/.openwiki/.env, keeping sensitive credentials off the repository.
OpenWiki includes built-in support for continuous integration workflows. The tool provides example CI configurations for both GitHub Actions and GitLab CI, enabling teams to automatically open pull requests or merge requests with documentation updates. This automation ensures that documentation stays current as the codebase evolves without requiring manual intervention.
The project welcomes community contributions and maintains a structured contribution process outlined in its CONTRIBUTING.md file. The maintainers intentionally keep pull requests tightly scoped to individual changes, closing PRs that bundle unrelated modifications and requesting they be split into separate submissions. This approach helps maintain code quality and review clarity as the project develops.
