Skills

Skills are folders of instructions, scripts, and resources that an agent loads dynamically to improve performance on specialized tasks. A skill teaches the agent how to complete a specific task in a repeatable way — writing a commit message in your team's style, drafting a PR description, polishing docs, and so on.

Each skill is self-contained in its own folder with a SKILL.md file holding the instructions and metadata. Browse the skills directory for working examples, or use the template as a starting point.

Repository layout

• ./skills — example skills you can use or learn from
• ./spec — the Agent Skills specification
• ./template — a minimal skill template

Install in Claude Code

Register this repository as a plugin marketplace:

/plugin marketplace add contextosai/skills

Then install a plugin. The production harness audit:

/plugin install harness-audit@contextosai-skills

Or the example skills (commit messages, PR descriptions, README polish):

/plugin install example-skills@contextosai-skills

Once installed, invoke a skill just by mentioning the task — for example, "audit my agent harness for production readiness" triggers harness-audit, and "write a commit message for my staged changes" triggers commit-message. You can also run harness-audit explicitly as /harness-audit.

Creating a skill

A skill is just a folder with a SKILL.md file containing YAML frontmatter and Markdown instructions:

---
name: my-skill-name
description: A clear description of what this skill does and when to use it.
---

# My Skill Name

Instructions the agent will follow when this skill is active.

## Guidelines
- Guideline 1
- Guideline 2

The frontmatter requires two fields:

• name — a unique identifier (lowercase, hyphens for spaces)
• description — what the skill does and when to use it; this is how the agent decides the skill is relevant, so cover both the what and the when

See spec/agent-skills-spec.md for the full specification, and template/SKILL.md for a starting point.

Available skills

Skill	Description
harness-audit	Production-readiness audit for an AI agent harness: scores 40 runtime controls against your agent's real code with file:line evidence and emits a fix queue.
commit-message	Write a conventional git commit message from staged changes.
pr-description	Draft a PR title and description from a branch's commits and diff.
readme-polish	Improve a README's clarity and structure without changing its meaning.

Validating

Every SKILL.md and the marketplace manifest are checked in CI on each push and pull request. Run the same check locally:

node scripts/validate-skills.mjs

Contributing

Contributions are welcome — see CONTRIBUTING.md for the skill format and the bar for inclusion, and please follow the Code of Conduct.

License

Released under the MIT License.