mehen

Rust-powered CLI for detecting heuristic source code metrics at scale: complexity, maintainability, lines of code, and more.

Designed for fast, deterministic analysis over large codebases, helping both human and AI engineers track how complexity evolves over time.

Install

npm install mehen

Or run without installing:

npx -y mehen --help
bunx mehen --help

Also available on PyPI:

uvx mehen --help

Supported Languages

• Python (.py)
• TypeScript (.ts)
• TSX (.tsx)
• Rust (.rs)
• Go (.go)

Usage

Analyze metrics for a directory:

npx -y mehen -m -p src

Export metrics as JSON:

npx -y mehen -m -p src -O json -o ./metrics

Other supported output formats: YAML, TOML, CBOR.

What Mehen Computes

• Cyclomatic complexity -- control flow complexity
• Cognitive complexity -- human-perceived complexity with nesting
• Maintainability Index -- overall maintainability score
• Halstead metrics -- volume, difficulty, effort, bugs prediction
• ABC metric -- assignments, branches, conditionals
• LOC family -- SLOC, PLOC, LLOC, CLOC, blanks
• NArgs / NOM / NExit -- arguments, methods, exit points
• NPA / NPM / WMC -- public attributes, public methods, weighted methods per class

CI Integration

mehen works well in CI pipelines. Here is a real-world example from a GitHub Actions workflow that computes metrics on pull requests, compares against the main branch, and posts a summary comment:

steps:
  - uses: actions/checkout@v5
    with:
      fetch-depth: 0

  - name: Set up Node.js
    uses: actions/setup-node@v5
    with:
      node-version: '22'

  # Run mehen on the PR branch
  - run: mkdir -p $HOME/mehen-json
  - run: npx -y mehen -m -O json -o "$HOME/mehen-json" -p src

  # Run mehen on main for baseline comparison
  - uses: actions/checkout@v5
    with:
      ref: main
      path: main
  - run: mkdir -p $HOME/mehen-json-base
  - run: npx -y mehen -m -O json -o "$HOME/mehen-json-base" -p main/src

  # Compare and comment on PR (using actions/github-script or similar)

The JSON output per file contains structured metric data that can be diffed across branches to surface regressions.

Platforms

Native binaries are provided for:

| OS | x64 | arm64 | |---|---|---| | Linux (glibc) | @mehen/linux-x64-gnu | @mehen/linux-arm64-gnu | | Linux (musl) | @mehen/linux-x64-musl | @mehen/linux-arm64-musl | | macOS | @mehen/darwin-x64 | @mehen/darwin-arm64 | | Windows | @mehen/win32-x64 | @mehen/win32-arm64 |

The correct binary is selected automatically at runtime.

Requires Node.js >= 18.

Links

• GitHub
• Issues
• PyPI package

License

MPL-2.0