Code Wiki generates interactive, auto-updating documentation from code repositories, with architecture diagrams, section-level exploration and a chat interface. Public repositories are browsable now; private repo support is coming soon.
Code Wiki is a documentation tool that turns a code repository into an interactive, browsable knowledge base. Instead of relying on hand-written docs that drift out of date, an AI agent reads the code and produces architecture overviews, API references and codebase insights, organized so a reader can move through the project section by section. The generated content is attributed to Gemini on the homepage, and the product sits under the Google domain.
The target audience is developers who need to understand unfamiliar code: engineers joining a large project, maintainers who want documentation without writing it, and teams evaluating open-source libraries before adopting them. The homepage showcases public repositories such as Go, Flutter, Kubernetes, React, the Model Context Protocol Python SDK and OpenClaw, each with its own generated wiki that can be opened immediately to see what the output looks like.
In practice, a reader picks a repository and lands on a structured wiki. Sections can be opened individually to see how a specific part works, and links connect descriptions back to the code, so an architectural overview leads to the exact service and a function summary leads to its definition in the repository. Diagrams convert the code into visuals of how components relate. A chat interface lets users ask questions in natural language, such as where a function is defined or how a piece of logic behaves. According to the homepage, documentation refreshes whenever a pull request is merged, which is the core promise against static docs.
Support for private repositories is marked as coming soon, with a notification sign-up instead of a live feature. That limits current use to the public repositories Code Wiki has already indexed. Among alternatives, it overlaps with README generators, docs-as-code platforms and general AI coding assistants, but its focus is narrower: continuously regenerated, navigable documentation tied to the source, with conversational lookup layered on top. Teams wanting wikis for proprietary code will need to wait for the private release.