Pattern miner: Add contribution-guidelines-checker archetype

[github-actions]

Mined pattern: Contribution Guidelines Checker

Upstream repositories and files mined:

• githubnext/agentics → workflows/contribution-guidelines-checker.md
• githubnext/agentics → workflows/contribution-check.md
• github/gh-aw → .github/aw/subagents.md (inline sub-agent dispatch pattern, corroborating evidence)

Recurring pattern found (2 upstream workflows):

Both upstream workflows check incoming pull requests against a repository's contribution guidelines (CONTRIBUTING.md and similar docs) and take a binary label-or-comment action:

• contribution-guidelines-checker.md triggers on pull_request: [opened, synchronize], searches CONTRIBUTING.md → .github/CONTRIBUTING.md → docs/CONTRIBUTING.md → README sections in order, then applies the contribution-ready label (max 1) or leaves one comment (max 1). It explicitly recommends a small model since it runs on every PR event.
• contribution-check.md performs the same guideline evaluation at scale by delegating each PR to an inline contribution-checker sub-agent (documented generally in .github/aw/subagents.md) and compiling results, since the target repo is external and has high PR volume.

Both workflows share the same observable shape: guideline discovery order, single-outcome labeling/commenting, and explicit DO NOT-merge/approve/close constraints — a distinct archetype from the existing pr-review (code correctness/security review) and content-moderation (spam/policy review of arbitrary content) archetypes.

Why the existing library did not already cover it:

• pr-review focuses on code correctness, security, and inline diff feedback — not documented process/contribution-guideline compliance.
• content-moderation never labels PRs ready/applies a binary compliance outcome tied to a specific document (CONTRIBUTING.md); it uses issues+pull-requests safe outputs for spam/policy review of arbitrary content, not a single-label-or-comment guideline check.
• No existing archetype references CONTRIBUTING.md discovery order or the small-model-for-high-frequency-PR-events tip.

What changed in patterns/:

• Added patterns/archetypes/contribution-guidelines-checker.json (curated: success_rate: null, count: 0, top_repos: []).
• Appended "contribution-guidelines-checker" to archetypes in patterns/manifest.json.
• Added the matching runtime rendering block (icon, capabilities, permissions, github_toolsets, body) to patterns/workflow-generation.json.

Validation results:

• npm test: 18 test files, 318 tests passed.
• npm run build: succeeded, new archetype JSON emitted to dist/patterns/archetypes/contribution-guidelines-checker.json.

Candidates deliberately left out:

• Orchestrator + inline sub-agent dispatch (contribution-check.md, .github/aw/subagents.md, .github/aw/multi-agent-research.md): only one upstream workflow demonstrates full sub-agent dispatch end-to-end; folded its guideline-discovery-order tip into this archetype instead of creating a separate archetype from a single example.
• Issue-arborist / sub-issue-closer parent-child issue linking: already covered by the existing issue-hierarchy-manager archetype.
• Duplicate-code-detector / large-file-simplifier assign-to-copilot remediation: already covered by the code-health-auditor archetype's "Assign findings to copilot" tip.
• Link-checker (broken link detection/fixing): only one upstream workflow found; no corroborating second example or explicit upstream doc recommendation, so it did not meet the two-workflow bar.

Generated by Pattern Miner · copilot · auto · 74.1 AIC · ⌖ 19.5 AIC · ⊞ 8.5K · ◷