Evidence-based engineering insights, straight from your git history and your code.

Developer tools that turn repositories into actionable intelligence grounded in academic research. Everything runs locally or inside your CI, and your data stays in your environment. All tools are focused in providing deterministic results and be cost-efficient, helping you burn less tokens for operations that can be done deterministally during runtime (and make better, informed decisions).

A Dart library and Model Context Protocol server with 30+ analysis tools that transform raw git history into insights on code quality, technical debt, delivery risk, and security compliance.

• 🧠 Research-backed algorithms: SZZ bug attribution, bus factor, logical coupling, entropy-based secret detection.
• 🔥 Bug hotspots, architectural drift, dependency freshness, velocity & burnout signals.
• 🤖 Plugs into Claude, Cursor, and other AI agents via MCP: npx -y @gbrandtio/rw-git-mcp.
• 📦 Or use it directly as a Dart API: pub.dev/packages/rw_git.

A GitHub Action that computes six language-agnostic quality metrics for every file changed in a pull request, with base-vs-head deltas, configurable quality gates, and a sticky PR comment.

• 📊 Cyclomatic, Cognitive & NPath complexity, ABC score, Halstead estimated bugs, Maintainability Index.
• 🌍 Works across Dart, JavaScript/TypeScript, Python, Go, Java, Kotlin, Rust, C/C++, and C#.
• 🚦 Fail the check when thresholds are violated (max-cyclomatic, min-maintainability, …).
• ⚡ Pre-compiled native binaries for fast CI runs.

Engineering decisions deserve better evidence than intuition. Our tools answer questions like "Where do bugs actually cluster?", "Who is the single point of failure on this codebase?", and "Did this PR make the code harder to maintain?" using peer-reviewed algorithms, computed on your machine.

Built for engineering leaders, platform teams, security auditors, and anyone who wants their AI agent to reason about repositories with real data.