热门技能
当前社区里最受关注、增长最快的技能。
agent-refactor-safety
youlianvr
'Use when the agent (or user) is about to: refactor/split a monolith, extract modules, merge two project versions, delete imports/functions, move changes between projects, or when after an edit "a line/feature disappeared" and it is unclear what broke. Covers: map before cut, import-and-lint loop, safe extraction order, compatibility facade, bind namespace + _LOCAL_NAMES guard, diff before copy, non-git merge protocol, regression protection (edits eat neighboring lines, verification AFTER the last edit). Do not use for incident debugging (debug-incident-protocol) and test writing (testing-discipline).'
browser-automation
youlianvr
Drive browsers and desktop GUIs — choose the right tool for the job. Covers agent-browser (standalone CDP), browser-skill (user's real browser via bsk), and computer-use (desktop GUI via cua-driver). Use when interacting with websites, web apps, native apps, or the desktop.
notebooklm
roomi-fields
This skill should be used when the user wants to query their Google NotebookLM notebooks for citation-backed, source-grounded answers, or manage notebooks, sources, and Studio content (audio, report, video, infographic, presentation, data table, flashcards, quiz, mind map). It drives the @roomi-fields/notebooklm-mcp engine — via the notebooklm MCP tools when they are available in the session, otherwise via its HTTP REST API — and covers Google login, citation formats, the daily-quota-aware batch/ingestion pattern, and source discovery.
cross-platform-gotchas
youlianvr
'Use when writing/editing cross-platform scripts and installers (Python/bash, Linux/macOS/Windows): platform.system() vs sys.platform, regex in raw strings (r"\.zip" matches literal backslash), FileNotFoundError from subprocess.run, package manager fallback (winget→scoop→choco), Windows registry PATH (fresh binaries not visible to process), bash = WSL stub on Windows (need Git Bash), single source of truth instead of copies. Verified on Windows install bug report.'
database-security
youlianvr
Use for authorized database security assessment covering PostgreSQL/MySQL/MSSQL/Mongo/Redis exposure, authz, UDF/command paths, and misconfiguration review.
writing-plugin-dotnet
AtomiCloud
Write or modify CyanPrint plugin code in C#. Use when the user asks to add validation rules, change plugin behavior, modify the entry point, run commands from a plugin, or mutate files. Covers entry point (CyanEngine.StartPlugin), native filesystem I/O (System.IO), and command execution (System.Diagnostics.Process). Plugins receive { Directory, Config } and use native OS operations.
organize-inbox
taika-izumi
"docs/inbox/ に置かれた未分類の情報を、フォルダ構成定義（docs/overview/folder-structure.md）の分類基準に照らして1件ずつ整理するスキル。移動だけでなく、複数分類への分割、既存ドキュメントへの追記統合も行う。ユーザーの明示実行、または start-work の inbox 検知からの提案で起動する。"
impl-validator
prateek
Validate whether an implementation matches its stated goal. Use this skill when a skill or agent wants a second opinion on its own output, when the user says "check this implementation", "validate what you did", "is this correct?", "review the output", or "did you do this right?". Also spawned automatically as a subagent by other skills (memory-bridge, daily-update) to self-check their outputs before presenting to the user. Returns a structured pass/warn/fail verdict with specific actionable issues.
api-design-reviewer
youlianvr
"Comprehensive REST API design review with automated linting, breaking-change detection, and design scorecards. Catches inconsistent conventions, missing versioning, and design smells before APIs ship. Use when reviewing a PR that adds or changes API endpoints, auditing an existing API for v2 migration, or establishing API standards for a team."
architecture-recovery
TheSoulGiver
Recover safely when architecture context is stale, invalid, missing, or contradicted by source evidence; inspect last-good context and architecture deltas without treating them as current truth. Do not use for an ordinary fresh architecture lookup.
db-first-search
youlianvr
Use BEFORE answering any question about workspace/project content: "where is this symbol defined", "who calls this function", "have we covered this before", "what's in the code", "what findings do we have on this topic" — instead of answering from memory or grep. Even when the db-tools MCP tools are available — load this skill first: answer ONLY via the database (search.py / MCP db-tools), NOT grep, NOT from memory. Also before writing/editing project code (index and understand via the database first). Covers: database selection (aggg2/sherpa-voice/research), build/freshness (build.py/--refresh), search (search.py FTS/--symbol/--calls/--inherits), findings (findings.py add/search/link), project map, when search isn't needed. Not for web research (web-research-camoufox) or questions unrelated to database content.
bt-menu-sales-reports
joecastelino
Run the Blackstone Toyota (BT, dealer 1249) Menu Sales scorecards — Daily Opened, Daily Closed, and Closed Month-To-Date — from the LIVE Tekion OpenAPI. The 4th sibling of the SCT/Kevin, BC/Ruben, and TOL/Sean pipelines. Use for any "BT menu report", "Blackstone Toyota menu sales", or Tony Garcia menu-sales report request. (Menu REBUILD work is a different skill — bt-tony-menu-rebuild.)
writing-resolver-typescript
AtomiCloud
Write or modify CyanPrint resolver code in TypeScript. Use when the user asks to change conflict resolution logic, modify merge strategies, handle file origins, or change resolution behavior. Covers entry point (StartResolverWithLambda), ResolverInput/ResolverOutput, ResolvedFile, and FileOrigin. Must ensure commutativity and associativity (sort, unique, deterministic ordering).
worklog-skillify
taika-izumi
"worklog-extract で採用された候補から、writing-skills 委譲で新規スキル作成または既存スキル拡張を行う。スコープ（汎用/プロジェクト固有/固有ルール）で成果物の配置先を振り分け、汎用パスをガイドライン配信元リポジトリ以外で実行しようとした場合は警告する。ユーザーまたは worklog-extract から起動。"
apollo-graphql
DonArtkins
"Apollo Client integration for Griot's GraphQL reads: single client, InMemoryCache typePolicies, colocated fragments, refetch discipline."
aggg2-mandatory-reads
youlianvr
Use when starting work in the imported AGGG2.0 project tree or answering questions about that project's content. In the Oper workspace, use the root AGENTS.md and current .agents/skills/ loader contract first; this skill is project-local guidance, not a replacement for Oper policy. Before acting inside an AGGG2.0 tree, read its available canonical context files in tiers (Tier A always and in full: AGENTS.md, CYCLE.md, CLAUDE.md; Tier B per task type: CAMOUFOX.md, DB-FIRST.md, AGENT-LSP.md, CODE-GRAPH.md, SETUP.md), then the task-type skill. Do not answer from memory or skip reading.
me:changelog-create
kbrdn1
"Méta-skill : génère un /changelog adapté à un projet (façon me:skill-create), basé sur le CLI changelog-generator, avec verbosité configurable gravée à la génération (+ override flags). Range dans changelogs/. Triggers: /me:changelog-create, créer un changelog par projet, générer la commande changelog du projet."
me:lettre
kbrdn1
Rédige une lettre de motivation de Kylian Bardini adaptée à une entreprise et une annonce précises, puis la rend en PDF A4 à la charte kbrdn.dev. Structure en quatre temps : preuve d'ouverture chiffrée, mapping exigence-par-exigence de l'annonce, traitement frontal de la mobilité Luxembourg et du « pourquoi vous », clôture avec disponibilité. Puise dans la même source de vérité que le CV (_candidature-kit/pool.fr.md) : jamais un fait inventé, et l'honnêteté sur ce qui manque est un argument, pas un aveu. À utiliser dès que l'utilisateur parle de « lettre de motivation », « cover letter », « lettre pour <entreprise> », « candidature à cette offre », colle une annonce en demandant une lettre, ou lance /me:lettre.
imessage
joecastelino
Send and receive iMessages/SMS via the imsg CLI on macOS.
cross-examine
zebbern
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, or mentions "grill me".
regression-modeler
zebbern
"Run regression analysis (OLS or logistic) on uploaded CSV/Excel data, generating coefficients, R², p-values, VIF, and plain-language interpretation. Triggered by requests for regression modeling, fitting data, testing significance, checking multicollinearity, or keywords like OLS, logit, coefficient, p-value, or R-squared."
yojam-config
prateek
Maintain Prateek's chezmoi-managed Yojam browser-router config. Use for desired mutations, disappearing or disabled rules, schema bumps, new rule UUIDs, or Mac onboarding. Edit the desired fragment, never the live config; installation belongs to chezmoi-management.
codebase-visualizer
kbrdn1
Transforme un dépôt en carte d'architecture isométrique interactive (HTML autonome, thème Claude Dark), publiée en artifact — blocs dimensionnés aux vraies LOC, arêtes dérivées du graphe d'imports mesuré, points de données animés, drill-down "go inside", trace pas-à-pas, mode diff entre deux révisions, plus un dossier technique mesuré (couches, cycles de dépendances, hubs, manifestes, chaîne d'outils, CI, binaires appelés). Déclencheurs — "visualise ce codebase", "carte isométrique du repo", "atlas du codebase", "diagramme visuel de l'architecture", "codebase atlas", /me:codebase-visualizer. Sert aussi à mettre à jour ou étendre une carte existante. Pas pour un schéma de mécanisme isolé dans une réponse en prose (SVG inline), pas pour du Mermaid, pas pour une maquette d'UI produit.
premortem
kbrdn1
"Exécute un premortem sur n'importe quel plan, lancement, produit, recrutement, stratégie ou décision. Suppose que ça a déjà échoué 6 mois plus tard et remonte dans le temps pour trouver tous les motifs. Produit un plan révisé avec les angles morts exposés. DÉCLENCHEURS OBLIGATOIRES : 'premortem ça', 'premortem mon', 'exécute un premortem', 'qu'est-ce qui pourrait tuer ça', 'stress-teste ce plan', 'qu'est-ce que je rate ici', 'trouve les angles morts'. DÉCLENCHEURS FORTS : 'qu'est-ce qui pourrait mal tourner', 'est-ce que je rate quelque chose', 'troue ce plan', 'où ça va casser', 'avocat du diable'. NE PAS déclencher sur des demandes simples de retour, des questions factuelles ou des demandes au Conseil LLM. DÉCLENCHER quand quelqu'un a un plan ou un engagement où le coût de se tromper est élevé."