CodeWiki MCP Server
Enables querying and analyzing Bitbucket repositories using Google CodeWiki's AI-powered code search and documentation.
Enables querying and analyzing GitHub repositories using Google CodeWiki's AI-powered code search and documentation.
Enables querying and analyzing GitLab repositories using Google CodeWiki's AI-powered code search and documentation.
Integrates with Google CodeWiki to provide AI-generated answers and documentation for code repositories via tools like listing topics, reading structures, and searching.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@CodeWiki MCP Server summarize the main functions in facebook/react"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
CodeWiki MCP Server
An MCP server that brings Google CodeWiki into your editor. Query any GitHub, GitLab, or Bitbucket repository and get AI-generated answers about the codebase — powered by Gemini.
Documentation · Release Notes · PyPI
Sample Conversation
Related MCP server: MCP Codebase Index
Quick Start
pip install codewiki-mcp
playwright install chromiumClient Setup
VS Code
Open Command Palette (Ctrl+Shift+P) → MCP: Add Server → Command (stdio) → enter codewiki-mcp.
Or add to .vscode/mcp.json:
{
"servers": {
"codewiki-mcp": {
"type": "stdio",
"command": "codewiki-mcp"
}
}
}Claude Desktop
Add to claude_desktop_config.json:
{
"mcpServers": {
"codewiki": {
"command": "codewiki-mcp"
}
}
}Docker
docker build -t codewiki-mcp .
docker run -it --rm codewiki-mcpTools
Tool | Description |
| Topics overview with previews |
| JSON table of contents |
| Full or section-specific docs (paginated) |
| Gemini-powered Q&A chat |
| Submit unindexed repos for indexing |
All tools accept repo_url as a full URL or owner/repo shorthand.
Documentation
Configuration, architecture, agentic AI guides, and more — see the full documentation.
License
MIT
Available Tools
5 toolscodewiki_list_topicsA
Retrieve the overview / available topics for a repository from Google CodeWiki.
Use this to discover what CodeWiki knows about a repo before asking
specific questions with codewiki_search_wiki.
Returns section titles with short content previews (not the full page).
For detailed content, call codewiki_read_contents with a section title.
Response size: typically 5–30 KB depending on the repository. Cached for 30 minutes — repeated calls for the same repo are instant.
Rate limit: max 10 calls per 60 s per repo URL. Duplicate concurrent calls are automatically deduplicated.
Args: repo_url: Full repository URL (e.g. https://github.com/microsoft/vscode-copilot-chat) or shorthand owner/repo (e.g. microsoft/vscode-copilot-chat). Bare keywords (e.g. 'vue') are auto-resolved with interactive disambiguation.
Inputs: repo_url (required). Returns: result
codewiki_read_contentsA
View documentation about a GitHub repository from Google CodeWiki.
Without section_title, returns the full wiki content (may be truncated).
With section_title, returns just that section's content.
Use codewiki_read_structure first to see available sections.
Pagination (when section_title is empty):
offset— section index to start from (default 0).limit— max sections per response (default 5). The response includeshas_moreandnext_offsetwhen more sections are available, so you can call again to continue.
Response size: 2–10 KB per section, 5–30 KB for paginated full page. Cached for 5 minutes — repeated calls are instant.
Rate limit: max 10 calls per 60 s per repo URL. Duplicate concurrent calls are automatically deduplicated.
Args: repo_url: Full repository URL (e.g. https://github.com/facebook/react) or shorthand owner/repo (e.g. facebook/react). Bare keywords (e.g. 'react') are auto-resolved with interactive disambiguation. section_title: Optional. Title (or partial title) of a specific section to retrieve. If empty, returns the full wiki. offset: Section index to start from (0-based, default 0). limit: Maximum sections to return (default 5, max 50).
Inputs: repo_url (required), limit, offset, section_title. Returns: result
codewiki_read_structureA
Get a list of documentation topics for a repository from Google CodeWiki.
Returns the table of contents / section structure as a JSON list so you
can choose which sections to read with codewiki_read_contents.
Recommended first step — call this before codewiki_read_contents
or codewiki_list_topics to discover available sections without
consuming many tokens.
Response size: typically 1–3 KB (lightweight JSON). Cached for 5 minutes — repeated calls are instant.
Rate limit: max 10 calls per 60 s per repo URL. Duplicate concurrent calls are automatically deduplicated.
Args: repo_url: Full repository URL (e.g. https://github.com/facebook/react) or shorthand owner/repo (e.g. facebook/react). Bare keywords (e.g. 'react') are auto-resolved with interactive disambiguation.
Inputs: repo_url (required). Returns: result
codewiki_request_indexingA
Request Google CodeWiki to index a repository that is not yet available.
Use this tool when codewiki_list_topics or codewiki_read_structure
returns a NOT_INDEXED error, indicating the repository has no
CodeWiki documentation yet.
This tool will:
Search for the repository on CodeWiki.
Click "Request repository" to open the submission dialog.
Fill in the GitHub URL and submit the request.
Return confirmation and next-step guidance.
Note: Google CodeWiki reviews requests and indexes repositories based on popularity and demand. There is no guaranteed timeline.
Args: repo_url: Full repository URL (e.g. https://github.com/owner/repo) or shorthand owner/repo (e.g. owner/repo). Bare keywords (e.g. 'vue') are auto-resolved with interactive disambiguation.
Inputs: repo_url (required). Returns: result
codewiki_search_wikiA
Ask Google CodeWiki a question about an open-source repository.
This uses the interactive chat feature powered by Gemini.
For reading wiki content directly, use codewiki_read_contents instead.
Results are cached for 2 minutes — repeated identical queries are instant.
Response size: typically 0.5–5 KB depending on the answer.
Rate limit: max 10 calls per 60 s per repo URL.
Args: repo_url: Full repository URL (e.g. https://github.com/microsoft/vscode-copilot-chat) or shorthand owner/repo (e.g. microsoft/vscode-copilot-chat). Bare keywords (e.g. 'vue') are auto-resolved with interactive disambiguation. query: The question to ask (required).
Inputs: repo_url (required), query. Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 5 tools
The tool purposes are mostly distinct, but `codewiki_list_topics` and `codewiki_read_structure` both return structural information, which could initially confuse an agent without careful reading of descriptions.
All tools follow a consistent `codewiki_verb_noun` pattern in snake_case, making the naming predictable and easy to understand.
With 5 tools covering discovery, reading, searching, and indexing requests, the count is well-scoped for the server's documentation-focused purpose.
The tool set covers all essential operations for a read-only documentation server: overview, structure, content retrieval, search, and handling missing repositories.
Maintenance
Related MCP Connectors
Ask any GitHub repository a question. Get source-backed answers.
Code intelligence for LLMs. Analyze, search, and retrieve code from any public git repository.
Ask how your codebase works. Hosted codebase intelligence for repos indexed on repowise.dev.
A cited wiki of your GitHub repo: search, read pages, find symbols and ask, with line citations.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceLeverages Google's Gemini AI with a 1M token context window for comprehensive codebase analysis, enabling intelligent code search, architecture analysis, and targeted improvement suggestions.MIT
- AlicenseNot gradedqualityDmaintenanceEnables semantic search across your codebase using Google's Gemini embeddings and Qdrant Cloud vector storage. Supports 15+ programming languages with smart code chunking and real-time file change monitoring.30 npm20MIT
- FlicenseBqualityDmaintenanceEnables analysis and querying of Git repository content (both public and private) through a unified tool that provides repository summaries, file structures, and full content optimized for LLM consumption.1-
- AlicenseAqualityFmaintenanceGemini-powered code review. Offers review-only and review-and-commit modes with integrated Gitleaks security scanning.2Apache 2.0