| name | creating-agent-skills |
| description | Create and audit Copilot skills with strong discovery metadata, compact workflows, and explicit completion checks. |
| license | MIT |
Creating Agent Skills
Use when
- Authoring a new
SKILL.md.
- Tightening skill discovery and workflow quality.
- Auditing skills for ambiguity or over-verbosity.
- Packaging reusable scripts, references, or templates with a skill.
Workflow
- Define expected outcome and triggers.
- Create
<skill-name>/SKILL.md; name must exactly match the parent folder.
- Write concise frontmatter and discovery-focused
description.
- Add minimal procedural steps, decision points, and completion checks.
- Move long details into referenced
scripts/, references/, or assets/ files.
Frontmatter
Required:
name: kebab-case, lowercase letters/numbers/hyphens only, max 64 chars, no namespaces, must match folder.
description: what the skill does and when to use it; max 1024 chars; primary routing signal.
Optional:
- Open Agent Skills spec:
license, compatibility, metadata, experimental allowed-tools.
- VS Code/Copilot:
argument-hint, user-invocable, disable-model-invocation, experimental context: fork.
Do not put version in SKILL.md; plugin/package versioning belongs in plugin.json or the distribution wrapper.
Layout
skill-name/
SKILL.md
scripts/ # optional executable helpers
references/ # optional focused docs
assets/ # optional templates/resources
Reference bundled files from SKILL.md with relative Markdown links so agents can load them on demand.
Guardrails
- Keep skills short and executable.
- Use "Use when..." trigger language with concrete keywords.
- Avoid embedding secrets or environment-specific credentials.
- Prefer skills for portable multi-file capabilities; use prompt files for one-off slash commands and custom agents for personas/tool restrictions.
Related docs