CodeWiki Bot is an AI-powered documentation and chat assistant that sits on top of a codebase. Instead of hunting through repos and outdated wikis, developers just ask — and get grounded answers, cross-repo dataflow diagrams, and architecture context on demand.
What it does
- Graph-RAG engine — combines a knowledge graph of the code with retrieval-augmented generation, so answers understand relationships between modules, not just keyword matches.
- Cross-repo dataflow & architecture diagrams — automatically generated from indexed code, spanning multiple repositories.
- Natural-language chat over the codebase — ask “where is X handled?” or “how does data flow from A to B?” and get a sourced answer.
- Semantic search across Azure DevOps repos — 12,000+ documents indexed and kept in sync.
Why Graph-RAG (not plain RAG)
Plain RAG retrieves similar text chunks. A graph layer adds the structure of the codebase — call relationships, data flow, module dependencies — so the bot can reason across files and repos instead of answering from isolated snippets. That’s what makes cross-repo diagrams and “how does this connect” questions possible.
Note: identifiers in the screenshot above are masked for confidentiality.