Adds gh-aw (GitHub Agentic Workflows) framework files plus an AI-driven daily hygiene-assessment workflow that scans main for: 1. Workaround patterns (CLAUDE.md forbids without approval) 2. Ungrounded retail-port code (no decomp citation) 3. Roadmap / phase / CLAUDE.md "currently working toward" drift 4. Test / build hygiene (warnings, test count regression) 5. Architecture drift (cross-layer references, WB imports outside Wb/) Output: one rolling GitHub issue per day, labelled ai+hygiene; the framework auto-closes the prior day's report. Engine: copilot (gpt-5.3-codex) — uses your Copilot subscription, no separate API key needed. Scaffolding from gh aw init: - .gitattributes (marks .lock.yml as generated) - .github/agents/ (dispatcher agent) - .github/mcp.json (MCP server config) - .github/workflows/aw.json (ghes:false — we target GitHub.com) - .github/workflows/copilot-setup-steps.yml - .vscode/settings.json (editor settings) Workflow: - .github/workflows/hygiene-assessment.md (human-authored source) - .github/workflows/hygiene-assessment.lock.yml (compiled artifact) Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
1 line
No EOL
63 B
Text
1 line
No EOL
63 B
Text
.github/workflows/*.lock.yml linguist-generated=true merge=ours |