Skip to main content
A skill is a structured Markdown file, typically named SKILL.md, that contains everything an AI coding agent needs to perform a specific task reliably.

The problem skills solve

AI coding agents like Claude Code, Cursor, and Copilot are powerful but generic. Without specific instructions, they:
  • Guess at your project’s conventions
  • Miss domain-specific requirements
  • Produce inconsistent results across sessions
  • Cannot learn from corrections
They give agents persistent, structured knowledge about how to do things in your specific context.

What a skill contains

A skill typically includes:
SectionPurpose
MetadataName, version, description, tags, target agents
InstructionsStep-by-step directions for the task
ConstraintsRules and guardrails the agent must follow
VerificationHow to check that the output is correct
ContextBackground information, conventions, examples
ExamplesSample inputs and expected outputs

Skills vs. prompts

PromptSkill
FormatFree textStructured Markdown with metadata
PersistenceOne-off, lives in chat historyVersioned file on disk
ReusabilityCopy-pasteInstall once, use across sessions
QualityVariesScored on 6 dimensions
SharingScreenshots, gistsExport, import, or publish

Skills across agents

Skills are designed to be agent-agnostic. The same skill can be installed into:
  • Claude Code via .claude/skills/ files or CLAUDE.md sections
  • Cursor via .cursor/skills/ files
  • GitHub Copilot via .github/copilot-instructions.md
  • Codex CLI via AGENTS.md
  • Gemini CLI via GEMINI.md
  • Amp via .amp/skills/ files
  • Goose via .goose/skills/ files
SkillCreator handles the format conversion automatically when you export.

Skill lifecycle

  1. Create. Write from scratch, generate with AI, or convert existing docs.
  2. Edit. Refine instructions, add constraints, include examples.
  3. Evaluate. Score quality and fix gaps.
  4. Install. Deploy to one or more agent runtimes.
  5. Use. Agents follow the skill during coding sessions.
  6. Patch. Update based on real-world performance.
Last modified on March 19, 2026