I built a better GitHub search that rank repos by what matters
GitHub search with README scoring, but Sourcegraph already does this better.
Audit any repository for AI-readiness.
Runs 66 weighted checks across repo hygiene, grounding docs, prompt templates, agent configs and AI dependencies, then spits out a 0–10 grade plus JSON so you can gate CI. Zero-deps npx install and an exit code threshold for CI are pragmatic touches that make it immediately usable. It could use better tuning/customization for false positives, but as an opinionated audit it saves the time of stitching together half a dozen ad-hoc checks.
Developers, repo maintainers, ML engineers and engineering teams preparing codebases for AI agents and integrations
GitHub search with README scoring, but Sourcegraph already does this better.
Agent-maintainable convention checks is novel but alpha stage needs real-world proof.
Single-page scorer for AI-ready repos when 'code quality' already covers this.
AST-parsed knowledge graphs across repos beat single-repo code chat tools.
Exposes hidden install costs and API key requirements that star counts completely hide.
Generates setup scripts for GitHub repos, but Devbox and Nix already solve this.