# Open-source DeepWiki

> Self-hosted generators that turn a code repository into a browsable, AI-written wiki.

These tools take a code repository and produce a browsable wiki written by an LLM: a table of contents, one page per topic with diagrams and links back to the source, and usually a chat over the code. They differ most in how the model gets its context. Some embed every file and retrieve chunks (RAG). Others let an agent list, grep and read files itself. When you choose, check four things. How many files the tool will really look at in a large repo. Whether citations point to real lines at a pinned commit. Whether you can use different models for the outline and the pages, including local ones. Whether a new commit updates only the changed pages or rebuilds the whole wiki. Running cost and setup weight also differ a lot, from one Python process to a .NET service with a database.


## Projects

- [AsyncFuncAI/deepwiki-open](https://llms-technical-reviews.com/p/deepwiki-open/) — FastAPI service and Next.js UI that embed a repo into a FAISS index, then generate a cited, Mermaid-heavy wiki with one prompt per page. (★18128, Python)
- [AIDotNet/OpenDeepWiki](https://llms-technical-reviews.com/p/opendeepwiki/) — ASP.NET Core service in which tool-calling LLM agents read a repo and write its wiki, mind map and translations into SQLite or PostgreSQL. (★3615, C#)


## Comparison questions

- [How is a repository ingested and chunked?](https://llms-technical-reviews.com/open-source-deepwiki/q/ingestion/) — Clone or local path; file filters; chunking strategy; supported hosts; large-repo limits.
- [How is retrieval (RAG) implemented?](https://llms-technical-reviews.com/open-source-deepwiki/q/retrieval/) — Embedding models; vector store; top-k; how retrieved code reaches the prompt; or agentic file reading instead of RAG.
- [How is the wiki structure (table of contents) determined?](https://llms-technical-reviews.com/open-source-deepwiki/q/structure/) — Prompt/agent that proposes sections and pages; inputs used (file tree, README); output format.
- [How are individual pages generated?](https://llms-technical-reviews.com/open-source-deepwiki/q/page-generation/) — Per-page prompts; source-file citations; diagrams (Mermaid); parallelism; caching and regeneration.
- [How are model providers configured?](https://llms-technical-reviews.com/open-source-deepwiki/q/providers/) — Supported providers; per-stage model selection; OpenAI-compatible endpoints; local models.
- [How is interactive Q&A / chat implemented?](https://llms-technical-reviews.com/open-source-deepwiki/q/qa/) — Chat over the repo; deep-research mode; streaming; conversation memory; MCP exposure.

Full comparison: https://llms-technical-reviews.com/compare/open-source-deepwiki/