Coding & Development

Use this skill when using AI to learn new technologies, navigate unfamiliar codebases, do interview prep, identify skill gaps, or build a personal learning system. Also use when helping someone learn rather than just doing the work for them. Trigger on keywords: learn, learning, understand, explain, teach me, how does X work, unfamiliar codebase, interview prep, skill gap, study, tutorial.

Use this skill for backend development — server logic, authentication, microservices, serverless functions, middleware, background jobs, caching strategies, and backend architecture patterns. Trigger on keywords: backend, server, API endpoint, middleware, authentication, authorization, microservice, serverless, background job, queue, caching, session.

Run work through a self-grill, prove, challenge, simplify, clean-up loop, installing its companion skills first. Use when the user says autopilot or principal mode, or hands over a multi-file change, a bug with no known cause, or work they will review later.

Planning and approval gates for state-changing work.

Grill the user relentlessly about a plan, decision, or idea. Use when the user wants to stress-test their thinking, or uses any 'grill' trigger phrases.

Finish real work with a Builder-Challenger pass. The building agent hands its finished work to a different model (Codex or Claude, called read-only) that challenges it under the Builder-Challenger Protocol, loops for up to 5 rounds (the Operator can extend once), and presents the objection ledger to the Operator (the human who owns the work) for go/no-go. Use when work is ready for completion approval, when AGENTS.md says "Finish with a challenge", or when asked to run BCP, challenge this, or get a second-model review. Skip for quick, obvious, easy-to-undo changes.

This skill should be used when the user asks about libraries, frameworks, API references, or needs code examples. Activates for setup questions, code generation involving libraries, or mentions of specific frameworks like React, Vue, Next.js, Prisma, Supabase, etc.

Use when about to edit, create, or delete an agent skill. Shared skills are sourced from the agent-skills repository; downstream-local skills stay with their owning repository.

코드가 다음 변경을 싸게 만드는지 본다. 기존 추상화를 우회한 중복, 레이어 위반, 변경 이유가 둘 이상인 단위, 호출자마다 같은 처리를 반복하게 만드는 API를 찾아 blocking과 suggestion으로 나눈다. "유지보수성 봐줘", "구조 리뷰", "설계 괜찮아?", "중복 없나", "머지 전에 점검", 에이전트 PR 리뷰, 커밋 전 셀프 체크, 기존 모듈 점검에 사용한다.

为16:9横屏AI漫剧创作、改编、审查和修订可生产剧本。用于将故事、小说或既有剧本转为90-180秒单集、可分镜和可连续制作的剧集文本；不负责资产图或视频生成。

Keep a project's dependencies up to date and free of known vulnerabilities. Finds outdated packages and security advisories, applies security fixes and safe patch/minor bumps, then takes major upgrades one at a time (reading the migration guide, fixing breakage, re-running lint/tests/build), and finishes with a branch, focused commits, and a PR. Works for npm/pnpm/yarn, Python (pip, uv, poetry), Rust (cargo), Go modules, and others through an OSV lockfile scan. Use this whenever the user mentions outdated packages, dependency updates, bumping versions, upgrading a library or framework to a new major, npm audit / pip-audit / cargo audit warnings, CVEs or GHSAs in dependencies, Dependabot or Renovate alerts, "is anything vulnerable", or routine dependency maintenance, even if they only ask to check and not to change anything.

Pre-indexed local code knowledge graph for navigating and understanding codebase structure (symbols, call paths, dependencies, blast radius) without a grep/read loop. Use when answering "how does X work", "what calls Y", managing CodeGraph daemons/indexes across Git branches or worktrees, or surveying an unfamiliar part of the active repository. Do not use for external/public repositories not yet indexed locally, or a repository with no .codegraph/ index — run codegraph init there first, or use deepwiki for public repos instead.

Verify every citation and quotation in an article draft against the real sources — cases via CourtListener + official reporters, scholarship via the user's Zotero library then the open web — and return a graded, footnote-by-footnote report with a "needs your attention" section first, plus a citation-form section (short cites per Bluebook R. 10.9/4.2 and italics/small caps per R. 2.1(a)) decided from the draft alone. Trigger when the user wants a draft cite-checked ("cite-check this", "check the citations/quotes in this draft", "run cite-check on <file>"). Input: a .docx with real Word footnotes (markdown/plain-text fallback). Not a full Bluebook linter — signal order, citation order and cite composition stay out.

Gather the full document set for a SCOTUS case being prepped for a podcast episode, class, or article — cert-stage filings, merits briefs, lower-court opinion, questions presented, and oral-argument transcript — and load them onto the user's reMarkable in a per-case folder. Manual trigger — use when the user says "prep <case>", "grab the briefs for <case>", "load up <case>", or similar.

>- Use when installing or using a Croct SDK in application code: rendering slot content, evaluating CQL queries, server or client fetching, prerendering, identifying users, app ID / API key env vars, testing, or debugging. Covers JavaScript (`@croct/plug`, `croct.fetch`, `croct.evaluate`), React (`@croct/plug-react`, `<CroctProvider>`, `useContent`, `<Slot>`, `useEvaluation`, `useCroct`), Next.js (`@croct/plug-next`, `fetchContent`, `cql`, middleware/proxy, `NEXT_PUBLIC_CROCT_APP_ID`, `CROCT_API_KEY`), Vue (`@croct/plug-vue`), Nuxt (`@croct/plug-nuxt`), Shopify Hydrogen (`@croct/plug-hydrogen`), PHP (`croct/plug-php`), Laravel, Symfony, and Drupal. Triggers: slot IDs like `home-hero@1`, `SlotContent`, "Croct is not plugged in", "initial content is required for server-side rendering", "Did you forget to configure Croct's middleware?", CORS errors.

Set up CountWatch monitoring for business counts (orders, sign-ups, payments, traffic): create an account with one API call, send or import counts, add a webhook and prove it with a test alert. Use when the user asks to monitor, watch or get alerted on a count or rate, or to find out when something stops, spikes or drops.

Vigila sismos y temblores cerca de la ubicación del usuario (por defecto Copiapó, Chile) usando USGS, EMSC y, si está configurado, GlobalQuake Server o el backend sismo-rt, y avisa en la conversación (y con notificación push) apenas se detecta uno. Úsala cuando el usuario diga "avísame si tiembla", "vigila los sismos", "¿hubo temblores cerca?", "¿tembló?", "último sismo", "GlobalQuake", "alerta sísmica" o pregunte por terremotos recientes o en curso.

This skill should be used when the user asks for an on-screen graphic, title or overlay - "bau eine Bauchbinde", "mach ein Lower Third", "blend meinen Namen ein", "mach eine Titelkarte", "setz einen Text drüber", "brauch eine Grafik dazu", "zeig da eine Zahl an", "mach ein Intro-Insert", "build a lower third", "add a title card", "put a graphic over this", "overlay some text", "animate this number". Covers the HTML template contract, Chrome headless frame rendering, alpha-preserving movie encoding, and the overlays block in project.json.

Use when looking up car repair procedures, torque specs, diagrams or service-manual steps. Scrapes free OEM manuals by vehicle, resolves the exact trim/engine, and cross-checks parts catalogs and field sources.

Compact the current conversation into a handoff document, saved to the repo's handoffs/ directory, for another agent or a fresh session to pick up.

Read this skill before the first step that uses the built-in browser, the browser pane inside the Claude desktop app (also called the in-app browser, the browser pane, Claude's browser, or \"your own browser\"), whose tools are named mcp__Claude_Browser__* when the session runs in the desktop app and mcp__remote-devices__Claude_Browser__* when a cloud session is linked to the person's computer; before those tools are turned on there may be a single enable__mcp__remote-devices__Claude_Browser tool instead. It covers the pane's persistent sign-ins, tabs and preview_start, reading pages as text, site approvals, what the pane cannot open, and what to do when it cannot be reached. It is not for Claude in Chrome (mcp__claude-in-chrome__* tools), which has its own skill, and it does not decide which browser to use.

Read this skill before the first step that uses Claude in Chrome, the browser extension whose tools are named mcp__claude-in-chrome__* (also called Chrome, the browser extension, or the external browser) and which acts in the person's real Chrome with their own sign-ins; before those tools are turned on there may be a single enable__mcp__claude-in-chrome tool instead. It covers loading the tools in one ToolSearch call, checking the person's open tabs and working in a new tab, site permissions, GIF recordings, console logs, dialogs to avoid, and when to stop and ask. It is not for the built-in browser (mcp__Claude_Browser__* or mcp__remote-devices__Claude_Browser__* tools), which has its own skill, and it does not decide which browser to use.

Coupe automatique d'un rush — supprime les blancs/silences, les hésitations (euh, hum…), les bégaiements (« je pense que je pense que ») et les prises ratées/répétées en gardant la dernière prise. Utiliser pour « enlève les blancs », « jump cuts », « supprime les répétitions », ou pour ajuster une coupe trop serrée/trop lâche.

Use when the user has a vague topic and wants an outline.