skills et frameworks communautaires pour créer des agents IA | AI Agent Skills
Collection de AI Agent Skills installables pour les workflows développeur community dans Claude Code, Cursor et Windsurf.
Cette collection s’adresse aux équipes qui construisent des agents IA avec l’écosystème communautaire. Elle regroupe des outils pratiques, des skills réutilisables et des frameworks pour le prototypage, le débogage, l’évaluation et l’intégration. L’objectif est de passer plus facilement de l’expérimentation à la production avec des workflows plus clairs et moins de solutions ad hoc.
Primary Install Bridge
Pick One Skill, Then Take the Install Path
This collection should not trap users in comparison mode or pretend to install the whole collection. Its job is to narrow the shortlist to one skill, then send the next click into installation, validation, and rollout. Installation happens on the skill path.
The Next Click Should Keep Narrowing, Not Reset Back To A Generic Directory
Once the install path is clear, move into the solution, CLI, or editorial surface that best matches this collection. That keeps platform, framework, and operations demand narrowing into a more verifiable high-intent journey.
Reviewed on 2026-04-17 against community workflow fit, installation clarity, operator handoff, and builder usefulness. This page is now positioned as an install-first community entry point instead of a broad community roundup.
We prioritize this page because community-intent users usually need a shortlist they can install, validate, and carry into real builder loops quickly.
Trust Signals
- Entries are chosen for practical community workflow value such as reuse, experimentation, debugging support, and installability.
- Selection favors tools with public documentation and clear setup paths that teams can validate before wider rollout.
- The page is curated for repeatable builder execution, not for vague community keyword coverage or generic repository popularity.
Grouping Logic
- Lead with tools that can enter a community-driven workflow without adding heavy setup ambiguity.
- Keep the shortlist compact enough for quick comparison while still covering reuse, debugging, experimentation, and workflow support.
- Use installation as the bridge from community discovery into validated daily execution.
Maintenance & Review
Last Reviewed
2026-04-17
Cadence
Re-check when install flow, maintainer posture, or community workflow relevance changes upstream; otherwise review monthly.
Maintained By
Killer-Skills editorial review within the recovery-focused authority queue.
Verification
Validate installability, operator clarity, workflow fit, and maintainer trust before retaining or adding an entry.
Execution Examples
How These Skills Work Together In Practice
Install one community skill first
Use this page when you want one community-tested skill that can enter real builder work without turning evaluation into another endless repo crawl.
1. Open the installation docs before opening more community repositories.
2. Choose one skill that best supports reuse, debugging, experimentation, or workflow speed.
3. Install it and verify the CLI write path, sync behavior, and first operator checkpoint.
4. Only after the base path works, expand the setup across the wider team workflow.
Reduce open-source selection drag
Treat the collection as an editorial filter when you need community-adjacent tools that improve builder confidence without overcomplicating the stack.
1. Check whether the tool has stable ownership and visible install guidance.
2. Review CLI behavior so operators know what will be written and synced.
3. Use one validated path before scaling it to more repos, flows, or teammates.
4. Document the chosen community workflow baseline after the first clean rollout.
Zustand state management guide. Use when working with store code (src/store/**), implementing actions, managing state, or creating slices. Triggers on Zustand store development, state management questions, or action implementation.
Data fetching architecture guide using Service layer + Zustand Store + SWR. Use when implementing data fetching, creating services, working with store hooks, or migrating from useEffect. Triggers on data loading, API calls, service creation, or store data fetching tasks.
Testing guide using Vitest. Use when writing tests (.test.ts, .test.tsx), fixing failing tests, improving test coverage, or debugging test issues. Triggers on test creation, test debugging, mock setup, or test-related questions.
Guide for adding keyboard shortcuts. Use when implementing new hotkeys, registering shortcuts, or working with keyboard interactions. Triggers on hotkey implementation or keyboard shortcut tasks.
The ultimate space for work and life — to find, build, and collaborate with agent teammates that grow with you. We are taking agent harness to the next level — enabling multi-agent collaboration, effortless agent team design, and introducing agents as the unit of work interaction.
Babysit a GitHub pull request after creation by continuously polling CI checks/workflow runs, new review comments, and mergeability state until the PR is ready to merge (or merged/closed). Diagnose failures, retry likely flaky failures up to 3 times, auto-fix/push branch-related issues when appropriate, and stop only when user help is required (for example CI infrastructure issues, exhausted flaky retries, or ambiguous/blocking situations). Use when the user asks Codex to monitor a PR, watch CI, handle review comments, or keep an eye on failures and feedback on an open PR.
Click-path-audit is a powerful AI agent skill that helps developers identify bugs in their UI by tracing every user-facing button and touchpoint. It benefits developers by providing a systematic approach to debugging, ensuring a seamless user experience.
Build a fully automated AI-powered data collection agent for any public source — job boards, prices, news, GitHub, sports, anything. Scrapes on a schedule, enriches data with a free LLM (Gemini Flash), stores results in Notion/Sheets/Supabase, and learns from user feedback. Runs 100% free on GitH...
The django-verification skill automates testing, linting, and security scans for Django projects, ensuring quality and security before release or PR. It benefits developers by saving time and reducing errors.
The foundation-models-on-device skill enables developers to build AI-powered features using Apple Intelligence on-device, generating or summarizing text without cloud dependency. It helps developers create privacy-preserving AI features with custom tool calling and snapshot streaming.
La compétence d'agent IA pour les modèles de serveur MCP permet aux assistants IA d'appeler des outils, de lire des ressources et d'utiliser des promts depuis votre serveur
La compétence d'agent IA Nutrient est un outil de traitement de documents qui utilise l'API de Nutrient pour convertir, extraire et rédiger des documents.
La sécurité Perl fait référence aux pratiques et techniques utilisées pour protéger les applications Perl contre les attaques et les vulnérabilités de sécurité
The project-guidelines-example skill provides a template for project-specific skills, leveraging Next.js, FastAPI, and Claude API. It helps developers manage complex projects and improve coding efficiency.
The santa-method AI agent skill provides multi-agent adversarial verification, helping developers ensure code accuracy and compliance by utilizing two independent review agents.
The search-first AI agent skill systematizes the research-before-coding workflow, assisting developers in finding existing tools and libraries before implementing custom code. This skill benefits developers by saving time and reducing redundant code.
Scan your Claude Code configuration (.claude/ directory) for security vulnerabilities, misconfigurations, and injection risks using AgentShield. Checks CLAUDE.md, settings.json, MCP servers, hooks, and agent definitions.
AI-assisted video editing workflows for cutting, structuring, and augmenting real footage. Covers the full pipeline from raw capture through FFmpeg, Remotion, ElevenLabs, fal.ai, and final polish in Descript or CapCut. Use when the user wants to edit video, cut footage, create vlogs, or build vid...
See, Understand, Act on video and audio. See- ingest from local files, URLs, RTSP/live feeds, or live record desktop; return realtime context and playable stream links. Understand- extract frames, build visual/semantic/temporal indexes, and search moments with timestamps and auto-clips. Act- transcode and normalize (codec, fps, resolution, aspect ratio), perform timeline edits (subtitles, text/image overlays, branding, audio overlays, dubbing, translation), generate media assets (image, audio, video), and create real time alerts for events from live streams or desktop capture.
Claude DevFleet est une compétence d'agent IA qui permet d'orchestrer des tâches de codage en parallèle, avec des rapports structurés et un suivi de progression
This AI agent skill streamlines C++ testing with GoogleTest and CMake, helping developers write and maintain robust tests. It automates test workflows for efficient code validation.
This AI agent skill provides a thorough Flutter/Dart code review, covering widget best practices, state management patterns, and clean architecture. Developers can improve their code quality and productivity with this skill.
The laravel-patterns skill provides production-grade Laravel architecture patterns for scalable and maintainable applications, benefiting developers with improved workflow and app performance.
Règles Distill est une compétence d'agent IA qui utilise des techniques d'apprentissage automatique pour extraire des principes transversaux de compétences et les convertir en règles.
Safety Guard is an AI agent skill that prevents destructive operations, ensuring safe coding practices for developers working on production systems. It protects against accidental data loss and system damage.
Use this skill when adding authentication, handling user input, working with secrets, creating API endpoints, or implementing payment/sensitive features. Provides comprehensive security checklist and patterns.
The springboot-verification skill automates testing and security scans for Spring Boot projects, helping developers ensure high-quality code. It integrates with Maven and Gradle for seamless workflow automation.
The customs-trade-compliance AI agent skill helps developers automate customs operations, ensuring lawful and cost-optimized movement of goods across borders. It assists with HS classification, duty calculations, and denied party screening, reducing the risk of penalties and seizures.
The energy-procurement skill helps senior energy managers optimize energy spend across multiple facilities. It analyzes tariff structures, identifies optimization opportunities, and evaluates demand charge mitigation strategies, leveraging AI coding for improved procurement decisions.
La planification de la demande d'inventaire est le processus de prédiction de la demande de produits et de gestion des stocks pour minimiser les ruptures de stock et les excédents de stock.
The production-scheduling AI agent skill helps senior production schedulers optimize job sequencing, line balancing, and changeover optimization, benefiting production management, planning, quality, and maintenance teams by maximizing throughput and meeting customer delivery commitments.
# Benchmark — Performance Baseline & Regression Detection ## When to Use - Before and after a PR to measure performance impact - Setting up performance baselines for a project - When users report "it feels slow" - Before a launch — ensure you meet performance targets - Comparing your stack agains...
La compétence d'agent de test de navigateur automatisé permet de tester l'interface utilisateur et l'interaction dans le navigateur de manière automatisée
This AI agent skill provides universal coding standards, best practices, and patterns for TypeScript, JavaScript, React, and Node.js development, helping developers maintain high-quality codebases. It assists in enforcing naming conventions, formatting, and structural consistency.
La recherche Exa est une compétence d'agent IA qui utilise le serveur Exa MCP pour effectuer des recherches neurales sur le web, le code et les entreprises.
Le système de conception de verre liquide d'iOS 26 offre une expérience utilisateur unique avec des effets de flou, de réflexion et de morphisme interactif pour SwiftUI, UIKit et WidgetKit.
La compétence d'agent IA regex-vs-llm-structured-text pour Claude Code fournit un cadre de décision pour choisir entre regex et LLM lors du parsing de texte structuré.
The tdd-workflow AI agent skill enforces test-driven development principles, ensuring comprehensive test coverage for developers. It streamlines coding workflows, reducing errors and increasing productivity.
Bun est un runtime JavaScript qui offre une alternative à Node.js avec des fonctionnalités de gestionnaire de packages, de bundleur et d'exécuteur de tests
Le modèle de cache de hachage de contenu utilise des hashes SHA-256 pour stocker les résultats du traitement de fichiers, améliorant l'efficacité et réduisant la charge de travail
Multi-platform content distribution across X, LinkedIn, Threads, and Bluesky. Adapts content per platform using content-engine patterns. Never posts identical content cross-platform. Use when the user wants to distribute content across social platforms.
Idiomatic Kotlin patterns, best practices, and conventions for building robust, efficient, and maintainable Kotlin applications with coroutines, null safety, and DSL builders.
La compétence de lentille de produit pour Claude Code est un outil de développement qui aide les développeurs à valider le 'pourquoi' derrière chaque fonctionnalité et à prendre des décisions éclairées sur leur produit.
Spring Security best practices for authn/authz, validation, CSRF, secrets, headers, rate limiting, and dependency security in Java Spring Boot services.
This AI agent skill enables developers to adopt Swift 6.2's concurrency model, ensuring single-threaded code by default and explicit background offloading. It helps resolve data-race safety compiler errors and optimizes app architecture.
Injection de dépendances basée sur les protocoles pour Swift, qui permet de simuler le système de fichiers, le réseau et les API externes pour des tests déterministes.
Notre compétence d'agent IA pour la gestion de la non-conformité de qualité aide les ingénieurs de qualité à gérer le cycle de vie complet de la non-conformité, de l'inspection à la disposition finale.
Carrier relationship management simplifies freight management for senior transportation managers, leveraging AI to optimize carrier onboarding, RFPs, and performance tracking.
La gestion des exceptions logistiques est le processus d'identification et de résolution des problèmes dans la chaîne d'approvisionnement, tels que les retards, les dommages ou les pertes de marchandises.
La compétence d'agent IA pour Claude Code fournit des modèles d'API d'Anthropic pour Python et TypeScript, couvrant l'API de messages, la transmission, l'utilisation d'outils et la vision.
The deployment-patterns skill enables developers to optimize web application deployment workflows using CI/CD pipelines, Docker, and health checks, ensuring production readiness and zero downtime.
dmux-workflows est une compétence d'agent IA qui permet l'orchestration de flux de travail parallèles dans Claude Code et d'autres environnements de développement
The e2e-testing AI agent skill helps developers build robust end-to-end test suites using Playwright and Page Object Model, streamlining their workflow and improving productivity.
Unified media generation via fal.ai MCP — image, video, and audio. Covers text-to-image (Nano Banana), text/image-to-video (Seedance, Kling, Veo 3), text-to-speech (CSM-1B), and video-to-audio (ThinkSound). Use when the user wants to generate images, videos, or audio with AI.
The laravel-verification skill automates environment checks, linting, static analysis, tests, and security scans for Laravel projects, benefiting developers by ensuring deployment readiness and security.
This AI agent skill provides comprehensive Perl testing patterns using Test2::V0, Test::More, and TDD methodology, helping developers ensure robust code quality and productivity.
This AI agent skill provides comprehensive Python testing strategies using pytest, TDD methodology, and best practices, helping developers write robust and maintainable code.
The continuous-learning skill automates pattern extraction from Claude Code sessions, saving developers time and effort. It identifies and saves useful patterns for future use, enhancing AI coding workflows.
This AI agent skill provides best practices for safe and reversible database migrations, benefiting developers working with PostgreSQL, MySQL, and common ORMs. It helps with schema changes, data migrations, and zero-downtime deployments.
The laravel-security skill provides comprehensive security guidance for Laravel applications, protecting against common vulnerabilities and ensuring secure deployment. Developers benefit from AI-driven security best practices for authentication, authorization, and more.
This skill provides Nuxt 4 app patterns for hydration safety, performance, and SSR-safe data fetching, benefiting developers with optimized routing and rendering decisions.
Regression testing strategies for AI-assisted development. Sandbox-mode API testing without database dependencies, automated bug-check workflows, and patterns to catch AI blind spots where the same model writes and reviews code.
The Architecture Decision Records skill helps developers capture and manage architectural decisions made during coding sessions. It auto-detects decision moments, records context, alternatives considered, and rationale, maintaining an ADR log for future reference.
Un ensemble de modèles de conception pour Compose Multiplatform et Jetpack Compose qui facilitent la création d'interfaces utilisateur partagées sur plusieurs plateformes
This AI agent skill provides structured concurrency patterns for Android and KMP projects, enabling developers to write efficient async code with Kotlin coroutines and Flows. It benefits developers by simplifying concurrent operations and error handling.
Spring Boot architecture patterns, REST API design, layered services, data access, caching, async processing, and logging. Use for Java Spring Boot backend work.
X/Twitter API integration for posting tweets, threads, reading timelines, search, and analytics. Covers OAuth auth patterns, rate limits, and platform-native content posting. Use when the user wants to interact with X programmatically.
L'outil de construction d'agents est une compétence de l'IA qui est utilisée pour améliorer l'efficacité des agents de l'IA en concevant et en optimisant les espaces d'action et les formats d'observation.
Clean Architecture patterns for Android and Kotlin Multiplatform projects — module structure, dependency rules, UseCases, Repositories, and data layer patterns.
La compétence d'agent d'intelligence artificielle pour ClickHouse est une capacité qui propose des modèles d'analyse et des techniques d'ingénierie de données pour optimiser les requêtes ClickHouse.
La compétence de l'agent IA pour JPA fournit des modèles de conception d'entités et de relations pour améliorer l'efficacité des applications Spring Boot
La compétence de modèles de serveur Ktor pour Claude Code propose un DSL de routage, une authentification, des WebSockets et plus pour améliorer la productivité.
SwiftUI architecture patterns, state management with @Observable, view composition, navigation, performance optimization, and modern iOS/macOS UI best practices.
The frontend-slides skill helps developers create engaging HTML presentations from scratch or by converting PowerPoint files, using AI-powered visual exploration for non-designers.
Une boucle continue d'agent est un modèle de conception qui permet la création d'agents autonomes avec des portes de qualité et des contrôles de récupération
The content-engine AI agent skill helps developers create strong, platform-native content systems for various platforms, adapting one idea into multiple formats. It benefits developers, marketers, and content creators by streamlining content creation and repurposing.
Enterprise-agent-ops operates long-lived agent workloads with controls for runtime lifecycle, observability, and safety, benefiting developers working with AI coding assistants like Claude Code.
La compétence d'agent IA pour les matériaux des investisseurs est une capacité qui aide à créer et à mettre à jour des documents pour investisseurs de manière cohérente et crédible.
La compétence d'agent IA pour Claude Code permet aux ingénieurs de travailler de manière plus efficace, en utilisant l'exécution eval-first et la décomposition pour améliorer la productivité.
AI-first engineering is a skill that helps developers design efficient processes, architectures, and code reviews for AI-assisted code generation, improving overall team productivity and code quality.
Draft cold emails, warm intro blurbs, follow-ups, update emails, and investor communications for fundraising. Use when the user wants outreach to angels, VCs, strategic investors, or accelerators and needs concise, personalized, investor-facing messaging.
Set up and manage the Sentry development environment using devenv. Handles fresh setup, updating existing environments, starting dev services, and troubleshooting. Use when asked to set up sentry, setup dev environment, get sentry running, start dev server, devenv setup, devservices not working, sentry wont start, or any development environment issue.
Sentry JavaScript frontend bug pattern review based on real production errors. Use when reviewing React/TypeScript frontend code for common bug patterns. Trigger keywords: javascript bug review, frontend errors, react error patterns, sentry frontend bugs.
Debug container agent issues. Use when things arent working, container fails, authentication problems, or to understand how the container system works. Covers logs, environment variables, mounts, and common issues.
Run initial NanoClaw setup. Use when user wants to install dependencies, authenticate messaging channels, register their main channel, or start the background services. Triggers on setup, install, configure nanoclaw, or first-time setup requests.
Read and extract text from PDF files — documents, reports, contracts, spreadsheets. Use whenever you need to read PDF content, not just when explicitly asked. Handles local files, URLs, and WhatsApp attachments.
Check out a GitHub pull request for review in this repo, either in the current directory or in a new isolated worktree at ../yaak-worktrees/pr-<PR_NUMBER>. Use when asked to run or replace the old Claude check-out-pr command.
Reviews Python code with an extremely high quality bar for Pythonic patterns, type safety, and maintainability. Use after implementing features, modifying code, or creating new Python modules.
Creates or updates README files following Ankane-style template for Ruby gems. Use when writing gem documentation with imperative voice, concise prose, and standard section ordering.
If This Page Is Close, Keep Narrowing With Adjacent Authority Pages
Do not reset back to the generic directory. Move sideways through these adjacent high-intent collections to narrow the shortlist toward the install path that best matches your team.
This collection should not keep users browsing forever. These three questions explain how to shortlist, install, and validate the next step.
Pour quels flux de travail ces collections sont-elles conçues ?
Ces collections sont conçues autour de l'automatisation de flux, processus, documents, données et piles de compétences réutilisables.
Quelle est la différence entre une collection et le répertoire principal ?
Le répertoire est meilleur pour la recherche directe, tandis que les collections regroupent des compétences complémentaires autour d'un flux complet.
Puis-je installer ces collections pour Claude Code ou Cursor ?
Oui. Les compétences de ces collections fonctionnent généralement dans Claude Code, Cursor, Windsurf et d'autres environnements via un flux d'installation unifié.
Additional Recovery Paths
Use These Additional Paths If You Need One More Step To Narrow The Decision
These are the supporting surfaces for this collection after the install direction is clear and the primary next paths have already narrowed the decision.
