AMS contribution-profile: inventory real-world eligibility signals across a diverse repo sample

[JSONbored]

Part of #6793.

Context

Before AMS can build a per-repo contribution profile, we need an accurate inventory of what real repos actually expose as eligibility signals — not an assumption, an audit of real examples.

Requirements

• Pick 8-10 real, diverse public GitHub repos (a mix: JSONbored's own gate-enabled repos, a few well-known OSS projects with different contribution norms, at least one repo with no explicit contribution docs at all) and document, per repo:
  • Its label taxonomy (via GET /repos/{owner}/{repo}/labels) and which labels (if any) read as eligibility/scope signals from their name+description alone.
  • Whether it has a CONTRIBUTING.md, what it says about issue/PR requirements (linked issue required? specific labels required? assignment rules?).
  • Whether it has a PR template (.github/PULL_REQUEST_TEMPLATE.md or similar) with an eligibility checklist.
  • Whether it has any AI-agent-facing docs (AGENTS.md, CLAUDE.md, .claude/skills/**, .cursor/rules, similar) that state contribution rules explicitly for an AI contributor.
• Produce a structured inventory (a table or per-repo breakdown) showing which signal types are present/absent across the sample, and how consistent or inconsistent the signal shapes are (e.g. do label descriptions reliably state point/eligibility semantics, or is it usually prose-only).

Deliverables

• A written inventory covering the sample repos, informing epic: AMS per-repo contribution-profile discovery (labels, docs, agent-facing eligibility signals) #6793's profile schema design (the next sub-issue). Research/documentation only — no code changes.

Expected Outcome

Whoever designs the ContributionProfile schema next has real evidence for which signal types are common/reliable enough to build extraction around, instead of guessing.

[linear-code]

JSO-4643