AI Skills Overview
GWS ships AI skills — playbooks installed into your AI agent (Claude Code, GitHub Copilot CLI, Codex CLI, Cursor, Amp) so you can drive your dev environment by typing slash commands inside the chat.
These are different from CLI commands:
| CLI commands | AI skills (this section) | |
|---|---|---|
| Where you run them | A terminal | Inside your AI agent's chat |
| How you invoke them | gws <command> | /gws-<skill> |
| What runs | The gws binary directly | The agent reads a SKILL.md playbook and calls multiple gws commands on your behalf |
| What they're for | Direct, scriptable operations | Higher-level workflows (setup, debug, health-check) where the agent reasons over output |
Installation
gws skills install
The CLI symlinks each skill into the directories your tools watch:
| Target directory | Picked up by |
|---|---|
~/.claude/skills/<skill> | Claude Code, GitHub Copilot CLI |
~/.agents/skills/<skill> | OpenAI Codex CLI, Cursor, Amp |
See gws skills for the full reference, including --force and --json.
What ships
| Skill | Slash command | Purpose |
|---|---|---|
| gws-setup | /gws-setup | Zero-config bootstrap — analyze repo, generate gws.json + Dockerfiles + K8s manifests, register, deploy, validate |
| gws-up | /gws-up | Start the environment with sanity checks and post-deploy verification |
| gws-status | /gws-status | Health-report all services, file-sync sessions, and URLs |
| gws-down | /gws-down | Tear down deployments cleanly (does not delete the cluster) |
| gws-debug | /gws-debug | 8-step debug runbook: doctor → pod signals → logs → exec → file-sync → URL probe → fix → escalate |
See also
- Zero-Config Setup with AI — the workflow
/gws-setupfollows gws skills— the installer- CLI Commands — the
gwscommands the skills call