YSCJRH

Audit and maintain consistency across AnswerLens public surfaces. Use when changing README copy, roadmap status, activation docs, release notes, Pages metadata, manual steps, or GitHub governance references, especially when latest release state, umbrella issues, or first-run entry points might drift out of sync.

Verify that AnswerLens onboarding and adoption paths stay copyable from demo to CI. Use when changing README first-run copy, demo walkthroughs, quickstart docs, GitHub Action docs, starter bundle examples, Pages entry points, or release-facing adoption guidance.

Maintain the AnswerLens activation funnel and public adoption narrative. Use when changing README copy, docs, Pages text, release/manual-step guidance, demo narrative, first-run story, or self-dogfooding surfaces for AnswerLens.

Productize AnswerLens first-run and starter-bundle adoption paths. Use when improving the fixture demo, real-site audit handoff, GitHub Action copy path, starter bundle, first-run story, or PR review artifacts.

Review AnswerLens public surfaces for unsafe claims, fake proof, premature npm install copy, secret leakage, raw provider payload exposure, or artifact-order drift.

Use when editing visitor-facing copy, navigation, ARIA labels, JavaScript-generated labels, focus behavior, mobile menu behavior, or Chinese-English bilingual UI on this personal website.

Use when reviewing or improving the public surfaces of this personal GitHub Pages site, especially homepage, Projects, Notes, Instrument Lab, navigation, links, responsive layout, metadata, or public-claim boundaries.

Use when Codex needs to explain, inspect, or validate Mirror's read-only deterministic Phase 1 public demo, including manifest, claims, evidence, eval summary, branch comparison, and public safety boundaries. Use for Mirror public demo onboarding, safe local demo checks, and claim/evidence review without model calls or filesystem-path artifact access.

Route 开始记录工作, 开始记录今天的工作, 停止工作并总结, 结束今天的工作并总结, 开始工作, 结束工作并总结, and 查看工作记录状态 to the local WinChronicle workday CLI without repository scanning or new capture surfaces.

Keep long-running Codex tasks scoped, validated, evidenced, and continuous.

Build and maintain a local, private digital double from interview answers and freeform self-description. Use when Codex needs to ask one high-value self-modeling question at a time, accept unstructured personal description, merge explicit corrections, and generate `profile.yaml`, `profile.md`, and a runnable double `SKILL.md` inside the local `doubles/slug-name/` folder without inventing memories or overstating certainty.