Linter, benchmarker, and CI gate for AI coding instruction files (CLAUDE.md, AGENTS.md, .cursorrules, etc.). Current stable version is v0.8.14, released with weekly cadence. Automatically detects dead references, token bloat, contradictions, stale instructions, broken links, and invalid skill metadata. Ships as a lightweight npm wrapper (~3 KB) that downloads a prebuilt native binary on first run; no Node runtime needed at execution time. Differentiators: zero-config startup, GitHub Action support, multi-format support (Claude, Copilot, Cursor, OpenAI), and built-in CI gating with regression detection.
npm install agenteval-cliNo compatibility data collected yet for this library.
Verified import paths — ran on the pinned version, not inferred.
Initializes, lints, and CI-gates AI instruction files using the CLI with common flags.
Use 'npx agenteval-cli' or 'npm install -g agenteval-cli'.
Download the binary manually from GitHub Releases and place it in your PATH, or use the Homebrew/curl install methods.
Use 'agenteval harvest --dry-run' for preview, but expect potential breaking changes in future versions.
Always pass decimal values, e.g., '--min-score 0.7' not '--min-score 1'.
Update to v0.8.13 or later, or manually kill the process with SIGTERM instead of closing terminal.
Use 'npx agenteval-cli' or install globally with 'npm install -g agenteval-cli'.
Set HTTP_PROXY/HTTPS_PROXY environment variables, or download binary manually from GitHub Releases and place it in PATH.
Use WSL on Windows, or use the experimental RISC-V build. Check GitHub Releases for platform availability.
Ensure npm global bin directory is in PATH. Run 'npm bin -g' to locate it.
No dependency data recorded yet.