# AI Agent Skills

Install the xtarterize skill for AI assistants (Claude Code, GitHub Copilot, Cline, etc.) to give them deep knowledge of the CLI, its JSON output, tasks, and best practices.

When installed, your AI assistant knows how to use `--format json`, parse task statuses, follow agent workflows, and avoid common mistakes.

## Install

```bash
npx skills add agustinusnathaniel/xtarter
```

## How it works

The skill activates when you ask about conformance, linting, CI, or any xtarterize task:

| You ask... | The agent... |
|------------|-------------|
| "Set up linting for this project" | Runs `check --format json`, parses task statuses, applies `init` or `add lint/biome` |
| "Check our conformance status" | Runs `check --format json`, reports which tasks are `skip` vs `new`/`patch` |
| "Add TypeScript strict mode" | Loads the task catalog, runs `add ts/strict --format json` |
| "What would change if I init?" | Runs `diff --format json`, presents structured file diffs |
| "Run diagnostics" | Runs `doctor --format json`, reports any `fail` diagnostics |
| "Restore a backup" | Uses `restore` to recover from `.xtarterize/backups/` |
**Tip:** The skill teaches agents to use <code>--format json</code> for every command and parse the structured output rather than terminal text.

## What's included

- All 8 CLI commands (`init`, `sync`, `diff`, `check`, `add`, `list`, `restore`, `doctor`)
- JSON output shapes for every command
- Parsing patterns — which JSON fields to inspect to decide the next action
- The 4 task statuses (`new`, `patch`, `skip`, `conflict`) and how to handle each
- Agent workflows for initialization, single tasks, diagnostics, and backups
- Anti-patterns — common mistakes agents should avoid

## Resources

- [CLI Reference](/xtarterize/guide/cli/overview/) — Full command reference
- [Conformance Tasks](/xtarterize/guide/tasks/overview/) — Task catalog
- [skills.sh](https://skills.sh) — Learn more about AI skills

[View CLI Reference →](/xtarterize/guide/cli/overview/)