L’interface est localisée ; les présentations et documents des Skills sont traduits automatiquement par IA dans votre langue. Le code, les commandes, chemins et termes techniques restent en version originale, et le document de l’auteur fait foi.
Explorer les Skills
Développement · SKILL.md

skill-name-here

skill-name-here

One to three sentences describing what this skill accomplishes, followed by key activation triggers. This field is the primary mechanism agents use to decide whether to activate the skill — it is read during discovery before the full body is loaded. Start with a verb. Include the most important "when to use" conditions inline. Max 1024 characters.

Independent GitHub snapshot · Automated classification · Original documentation governs

Documentation de la Skill

GitHub

name: skill-name-here description: > One to three sentences describing what this skill accomplishes, followed by key activation triggers. This field is the primary mechanism agents use to decide whether to activate the skill — it is read during discovery before the full body is loaded. Start with a verb. Include the most important "when to use" conditions inline. Max 1024 characters. license: MIT allowed-tools: Read Write Edit Bash Grep Glob metadata: author: Your Name version: "1.0" domain: general # match the parent directory under skills/ complexity: intermediate # one of: basic, intermediate, advanced language: multi # one of: R, TypeScript, Python, Docker, Rust, multi tags: tag1, tag2, tag3 # 3-6 lowercase tags for discovery; include the domain

Skill Title (Imperative Verb Form)

One paragraph: what this skill accomplishes and the value it provides.

When to Use

  • Concrete scenario where an agent should activate this skill
  • Another trigger condition
  • A third use case

Inputs

  • Required: Description of required input
  • Required: Another required input
  • Optional: Optional input with default (default: value)

Procedure

Step 1: Action Title

Context sentence explaining what this step accomplishes.

# Concrete command or code the agent can execute
example_command --flag value

Expected: What success looks like. Be specific -- file created, output matches pattern, command exits 0.

On failure: Recovery action. What to check, what to retry, when to abort.

Step 2: Next Action

Context for this step.

next_command

Expected: Specific success indicator.

On failure: Recovery action.

Validation

  • First verification check
  • Second verification check
  • Third verification check

Common Pitfalls

  • Pitfall name: Explanation and how to avoid it
  • Another pitfall: Explanation and how to avoid it

Examples (Optional)

Short worked examples showing the skill in action. Keep inline examples brief; move extended or multi-variant examples to references/EXAMPLES.md.

Related Skills

  • related-skill-name -- how it relates to this skill
  • another-skill -- how it relates
Skills enregistrées