Analyze project codebase health and vibe-coding risk with a 0-100 scoring system across 12 dimensions (plus a conditional AI App Surface dimension for LLM applications), centered on AI-as-maintainer concerns. Deterministic per-dimension scoring with a standing-deduction ledger that automatically re-verifies prior findings on every scan. Outputs a structured diagnostic report with scores, findings, paired Refactoring Risk + AI Extension Risk metrics (Drift Watch in the high-score regime), static dependency graph, and a ready-to-use remediation prompt (interactive or autonomous). Use this skill whenever the user asks to audit, analyze, score, or health-check their project, mentions "vibe coding risk", "code quality", "project health check", "risk analysis", "dependency graph", or wants to understand their codebase structure before refactoring — even if they don't explicitly say "vibe coding" (in any language).
Inspect commits since the last published release and decide whether to cut a new version of create-eikon-react (and at what semver bump), then draft the release commit + tag + npm publish steps. Use when the user asks to "check what's been released since X", "should I bump the version", "make a release", "发新版本", "提交一个新版本", or similar. Targets this monorepo's three-package layout (create-eikon-react public, @eikon/preview + @eikon/react private).