• Code Wiki is a web-based platform developed by Google.
• It automatically generates and maintains interactive documentation wikis for code repositories.
• The platform continuously scans repositories to regenerate structured documentation (including architecture and class/sequence diagrams).
• It features an integrated chat interface powered by the Gemini model to answer user questions by referencing the actual code.
• Code Wiki currently supports public repositories and plans to extend support to private repositories through a CLI integration.
Automated generation of documentation for code repositories after each change.
Context-aware chat assistant that answers questions about the code, tied directly to code files and definitions.
Interactive diagrams (architecture, class, sequence) built from the codebase.
Hyperlinked documentation sections: explanations linked to actual classes/functions.
Instant onboarding support: new contributors can understand large codebases more quickly.
Public-repo support initially, with upcoming private-repo support via the Gemini CLI plugin.
Regeneration on each commit/change to keep docs synchronized with code.
Indexing of large open-source projects (hundreds already documented).
Free preview (public open-source) offering; enterprise/paid model for private/internal use expected.
Emphasis on reducing the time developers spend reading code and increasing productivity.
What exactly does Code Wiki do?
It takes a code repository and automatically builds a structured documentation “wiki” that stays updated as the code changes, with diagrams and a chat interface to ask questions about the code.
Can I use it for my private company codebase?
At present, Code Wiki supports public repositories; support for private repos via a CLI (Gemini CLI extension) is coming soon.
Does the documentation remain stable or does it change with every commit?
Yes, one of the features is that the documentation regenerates after changes to ensure it remains in sync. Some users raise concerns about too-frequent changes impacting readability.
How reliable is the AI-generated documentation?
While the system can significantly speed understanding, Google acknowledges the AI can make mistakes and recommends verification of critical parts.
What types of diagrams and outputs are generated?
The platform generates architecture diagrams, class and sequence diagrams, linkable sections tied to code files/classes, and a chat UI overlay.