AI-only repository intelligence

Find useful AI projects with evidence, not hype.

Models, agents, MCP, RAG, inference, evals, multimodal, coding, local AI, safety, and datasets. Rankings are independent and not affiliated with GitHub.

Independent recommendation index

Repository signals

Loading…

    Published method

    A bounded 100-point score

    Momentum 40: min(40, 12 × ln(1 + observed star gain)). Missing history contributes zero and is labeled Baseline collecting.

    Freshness 15: linear decay over 730 days since creation. Discovery 10: min(10, 2 × log10(1 + stars)), deliberately capped. Maintenance 10: last push ≤7/30/90/180 days earns 10/7/4/1. License 5: an observed license is present. AI relevance 20: 5 points per matched deterministic topic or substantive metadata phrase, capped at 20.

    Score ties resolve by observed gain, then full repository name. Archived, forked, private, and weak “AI” marketing matches are excluded.

    Source & independence

    Dynamic repository facts come only from GitHub’s official REST Search API. Editorial category, relevance evidence, reason, and score are AI Repo Radar classifications. GitHub does not produce or endorse this ranking. Collection is scheduled every six hours; 7-day movement appears only after an eligible stored baseline exists.