Maîtrisez les outils et techniques d'IA
161 compétences ia
Skill agent reutilisable issu de l ecosysteme skills.sh. vercel-react-native-skills from vercel-labs/agent-skills: React Native and Expo best practices for performant mobile apps across list rendering, animations, navigation, and native modules.
Skill agent reutilisable issu de l ecosysteme skills.sh. minimal-run-and-audit from lllllllama/ai-paper-reproduction-skill: minimal-run-and-audit packages reusable procedural knowledge for agents.
Skill agent reutilisable issu de l ecosysteme skills.sh. grill-with-docs from mattpocock/skills: grill-with-docs packages reusable procedural knowledge for agents.
Skill agent reutilisable issu de l ecosysteme skills.sh. env-and-assets-bootstrap from lllllllama/ai-paper-reproduction-skill: env-and-assets-bootstrap packages reusable procedural knowledge for agents.
Skill agent reutilisable issu de l ecosysteme skills.sh. tdd from mattpocock/skills: Test-driven development with vertical slices, behavior-focused tests, and incremental red-green-refactor cycles.
Skill agent reutilisable issu de l ecosysteme skills.sh. paper-context-resolver from lllllllama/ai-paper-reproduction-skill: paper-context-resolver packages reusable procedural knowledge for agents.
Skill agent reutilisable issu de l ecosysteme skills.sh. azure-cost from microsoft/azure-skills: azure-cost packages reusable procedural knowledge for agents.
Skill agent reutilisable issu de l ecosysteme skills.sh. improve-codebase-architecture from mattpocock/skills: Analyze codebases for architectural friction and propose module-deepening refactors as testability improvements.
Skill agent reutilisable issu de l ecosysteme skills.sh. lark-contact from larksuite/cli: lark-contact packages reusable procedural knowledge for agents.
Skill agent reutilisable issu de l ecosysteme skills.sh. simple from roin-orca/skills: Lightweight brainstorming framework for fast, focused design decisions before implementation.
Skill agent reutilisable issu de l ecosysteme skills.sh. lark-minutes from larksuite/cli: lark-minutes packages reusable procedural knowledge for agents.
Skill agent reutilisable issu de l ecosysteme skills.sh. sleek-design-mobile-apps from sleekdotdesign/agent-skills: AI-powered mobile app design tool with REST API for creating projects, describing designs in plain language, and rendering screens.
Skill engineering
A skill is not just a long prompt. It is a packaged procedure: activation metadata, task instructions, optional scripts, reference files, examples, and validation steps. Good skill pages should explain what the skill does, when it activates, what resources it loads, and how users can verify the result.
What changed
Packaged procedure
Claude describes skills as directories with instructions, scripts, and resources, with a SKILL.md file defining activation and behavior. That makes skills suitable for team workflows such as SEO audits, spreadsheet analysis, slide generation, brand writing, and code review.
Progressive disclosure
Claude’s documentation describes a three-step pattern: lightweight metadata at startup, full instructions only when the task matches, and extra resources only when needed. This directly addresses context-window overload.
Runtime integration
OpenAI’s Agents SDK includes a skills capability that mounts skills into an auto-discovery root inside a sandbox and supports lazy loading. In practice, that turns skills into portable operating procedures an agent can discover during work.
Popular starting points
Explore AI skills by learning goal: find an AI skills course, learn AI skills for free, compare structured programs, understand AI skills in demand, and map skills to AI engineer requirements.
A practical starting point for prompt engineering, AI tools, agent workflows, data skills, evaluation, and applied workplace AI literacy.
Best for users who need a staged learning path from beginner foundations to practical projects and advanced agent workflows.
Useful for career-focused readers comparing workplace AI skills, AI engineer skills required, and portfolio projects that prove ability.
Explore AI skills learning paths
The page should combine learning, certification, role readiness, and agent-skill workflows instead of treating skills as generic tutorials.
These queries need a staged path: foundations, prompt skills, AI tools, agent workflows, data handling, evaluation, and portfolio projects.
These searches point to structured programs. The page can cover how to compare AI skills passport, Microsoft AI Skills Fest, and similar learning programs without overclaiming affiliation.
These users care about employability. Add sections about AI engineer skills required, workplace AI literacy, and evidence of skill through projects.
Skills vs prompts vs agents
These three layers should work together. A prompt instructs one task, a skill packages repeatable expertise, and an agent executes across tools and time.
What makes a skill worth learning
Related AI skills topics
Use these topics to move from basic AI skills into courses, free learning paths, career skills, engineering requirements, and agent skill workflows.
Skill FAQ
Explain the task it handles, the trigger conditions, the workflow, required tools or files, expected output, quality checks, and common failure cases. That is more useful than listing broad benefits.
Projects and custom instructions are broad background context. Skills are task-specific procedures that load dynamically, which makes them better for specialized work that should not always be in the context window.
High-value skills encode work that happens often and has a review standard: SEO content audits, research briefs, spreadsheet modeling, slide creation, code review, data cleaning, customer support QA, and brand-compliant writing.