Repolode
Submit
tirth8205

tirth8205/code-review-graph

A local code knowledge graph that gives AI coding tools precise review context over MCP and a CLI.

View on GitHubWebsite
Stars
32,255
Forks
3,055
Open issues
176
Last commit
3 days ago

What is it?

AI coding assistants often re-read large parts of a codebase to review a change. code-review-graph parses a repository with Tree-sitter into a graph of functions, classes, imports, calls, inheritance and tests, stores it in a single local SQLite file, and keeps it updated incrementally as files change.

When an assistant reviews a change, it queries the graph for the blast radius: the callers, dependents and tests that could be affected. The MCP server exposes about 30 tools for this, along with architecture overviews, execution flows, dead-code detection and semantic search. The same analysis runs as a composite GitHub Action that posts risk-scored comments on pull requests.

The project supports a wide range of languages and frameworks, including Python, TypeScript, Go, Rust, Java and PHP/Laravel, and it installs into common AI coding tools such as Claude Code, Cursor, Codex and Windsurf. Graph data stays on the local machine unless you opt into cloud embeddings.

Key features

  • 01Builds a Tree-sitter-based code graph and updates it incrementally by file hash
  • 02Computes blast radius to show which callers, dependents and tests a change affects
  • 03Exposes about 30 MCP tools and 5 workflow prompts to AI assistants
  • 04Runs as a GitHub Action that posts risk-scored PR review comments
  • 05Stores everything in one local SQLite file with no external service required
  • 06Supports custom languages through a languages.toml file

Use cases

  • Letting Claude Code or Cursor review a pull request without reading the whole repository
  • Finding which tests and callers a change to a shared function could break
  • Gating merges in CI with risk-scored review comments on pull requests
  • Getting an architecture overview of a large unfamiliar codebase

Topics

#code-review#mcp#static-analysis#tree-sitter#knowledge-graph#claude-code#llm#ai-coding#claude#graphrag

FAQ

What is code-review-graph?
A local code knowledge graph that gives AI coding tools precise review context over MCP and a CLI. AI coding assistants often re-read large parts of a codebase to review a change. code-review-graph parses a repository with Tree-sitter into a graph of functions, classes, imports, calls, inheritance and tests, stores it in a single local SQLite file, and keeps it updated incrementally as files change.
Is code-review-graph free to use? What license does it have?
code-review-graph is open source under the MIT license. Check the license terms before using it in commercial products.
Is code-review-graph still maintained?
Actively maintained. The last commit to code-review-graph was on Oct 16, 2026.
What can I use code-review-graph for?
Letting Claude Code or Cursor review a pull request without reading the whole repository Finding which tests and callers a change to a shared function could break Gating merges in CI with risk-scored review comments on pull requests Getting an architecture overview of a large unfamiliar codebase
Where can I get code-review-graph?
The source code and installation instructions are on GitHub: https://github.com/tirth8205/code-review-graph

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects