#skill
26 pages- BabysitStay on one pull request until it is merge-ready, fixing the highest-priority blocker and waiting when nothing is actionable.2026-10-04
- CalldiffPaste who-calls-whom diffs from calldiff into plans and pull requests when a change rewires call flow.2026-10-04
- CavemanUltra-terse replies that drop filler while keeping full technical correctness.2026-05-14
- DeslopRemove AI-generated code slop from a branch diff — unnecessary comments, defensive checks, any-casts, and nested code inconsistent with the surrounding codebase.2026-06-13
- Find SkillsDiscover and install agent skills from the open skills ecosystem at skills.sh using the npx skills CLI.2026-06-13
- Fix CIFind the first actionable failure in a PR's CI checks, apply the smallest safe fix, and loop until all checks are green.2026-10-04
- Fix Merge ConflictsResolve merge conflicts non-interactively with correctness-first edits, validate the build, and summarize resolution decisions.2026-10-04
- Frontend DesignBuild distinctive, production-grade frontends instead of generic AI UI.2026-06-13
- Git ConventionsConventional commit format and branch naming rules used as a reference when writing commit messages or advising on version control practices.2026-10-04
- Grill MeRelentlessly interview a plan or design, one question at a time, until every branch of the decision tree is resolved.2026-06-13
- HandoffCompresses a working session into a document another agent can use to resume without re-reading the whole chat.2026-05-14
- HumanizerRemove AI tells from prose so it reads like a person wrote it.2026-06-13
- Improve Codebase ArchitectureSurfaces deepening opportunities so shallow modules become deep, testable seams anchored in domain language and recorded decisions.2026-05-14
- AI SkillsReusable operating manuals for LLM agents.2026-10-04
- Iterate BrowserDebug frontend issues autonomously using browser tools and debug traces until the fix is verified — without asking the user to test manually.2026-10-04
- PR DescriptionWrite a pull request body a reviewer can predict the diff from, scaled to the size of the change.2026-10-04
- QAExercise a change the way a user would, capture evidence, and report a verdict instead of treating a type check as done.2026-10-04
- Recomposing CommitsRestructure messy branch history into logical commits using worktree isolation and a mandatory review gate.2026-06-13
- Request Refactor PlanInterview the user about a refactor, explore the codebase, break the work into tiny commits, and file the plan as a GitHub issue.2026-05-10
- Show MeExplain the current topic with the smallest diagram, sketch, or focused HTML artifact that makes the point.2026-10-04
- SimplifySimplify scoped code with parallel read-only review subagents, then apply targeted cleanup fixes without a large refactor.2026-06-13
- Split React ComponentsExtract inlined sub-components, icons, and data constants out of bloated React files into single-responsibility files.2026-05-10
- Thermo-Nuclear Code Quality ReviewAn unusually strict code review focused on structural ambition, abstraction quality, and deleting complexity rather than rearranging it.2026-06-13
- Weekly ReviewSynthesize shipped commits from the past 7–10 days into a short categorized summary with highlights by bug fix, tech debt, and new functionality.2026-06-13
- You Might Not Need an EffectAudit React code for useEffect anti-patterns per react.dev guidelines and optionally apply fixes in place.2026-06-13
- Zoom OutAsks for a higher-level map of modules, callers, and domain vocabulary before diving into unfamiliar code.2026-05-14