Menu
SkillsMP has collected 18 skills from Dat071104/ai-agent-workspace-pack. Open a skill to review its source and details.
Showing 18 of 18 collected skills.
Use this when the user wants to implement a new feature, endpoint, module, or behavior change in an existing codebase. Plans the change, confirms placement and business rules, then writes code in one serialized lane with tests. Not for bug fixes or cleanup.
Use this when the user wants a strong Codex, Cursor, Claude Code, Windsurf, phase, audit, harness, bug-fix, production-readiness, or handoff prompt. Use the readiness gate before large prompts.
Use this for git safety, public repo cleanliness, forbidden tracked files, generated artifacts, release readiness, and safe staging. Do not use for feature work or broad code cleanup.
Use this when the user reports a bug or failed behavior. Verify first, classify bug vs misunderstanding vs expected behavior vs feature request, then propose a minimal fix before editing.
Use this only for high-risk cleanup or refactoring. Requires warning, clean git state, recovery branch/tag, audit first, batch plan, one batch at a time, tests after each batch, and confirmation before continuing.
Use this for session handoffs, final reports, continuity summaries, next-session prompts, and milestone delivery notes. Do not use for code changes or bug fixing.
Use this for full audits, scoped audits, UX/performance audits, and production-readiness checks. This team tests and reports only; it does not fix by default.
Use this when the user wants to implement a new feature, endpoint, module, or behavior change in an existing codebase. Plans the change, confirms placement and business rules, then writes code in one serialized lane with tests. Not for bug fixes or cleanup.
Use this when the user wants to analyze a new project idea, compare architecture options, decide a roadmap, or clarify scope before writing files. Do not use for bug fixing, audit-only work, or code cleanup.
Use this for a senior advisor/mentor overview of a whole project. Reads the context card and idea, then prioritizes improvement opportunities by business value and estimates difficulty. Read-only; recommends and routes, does not change code. Use analyze-team…
Use this for a senior advisor/mentor overview of a whole project. Reads the context card and idea, prioritizes improvement opportunities by business value, and estimates difficulty. Read-only. Use analyze-team for options on a single decision.
Use this when the user wants to analyze a new project idea, compare architecture options, decide a roadmap, or clarify scope before writing files. Do not use for bug fixing, audit-only work, or code cleanup.
Use this when the user reports a bug or failed behavior. Verify first, reason across multiple root-cause hypotheses and fix directions, then recommend one minimal fix before editing. Can run parallel bug_hunter subagents for hard bugs.
Use this only for high-risk cleanup or refactoring. Requires warning, clean git state, recovery branch/tag, audit first, batch plan, one batch at a time, tests after each batch, and confirmation before continuing.
Use this for session handoffs, final reports, continuity summaries, next-session prompts, and milestone delivery notes. Do not use for code changes or bug fixing.
Use this when the user wants a strong Codex, Cursor, Claude Code, Windsurf, phase, audit, harness, bug-fix, production-readiness, or handoff prompt. Use the readiness gate before large prompts.
Use this for git safety, public repo cleanliness, forbidden tracked files, generated artifacts, release readiness, and safe staging. Do not use for feature work or broad code cleanup.
Use this for full audits, scoped audits, UX/performance audits, and production-readiness checks. Can detect and run the project's real test harness (Playwright, Jest, Vitest, pytest) after approval. Reports only; does not fix by default.