An Agent Skill that teaches your coding agent how to drive the Coolify CLI — deploy apps, manage databases & one-click services, configure servers, sync env vars, schedule backups, and stream deployment logs on Coolify Cloud or any self-hosted instance.
The skill wraps the official
coolify-cliand is distilled from its upstreamllms-full.txt— the full machine-readable command catalog.
Without it, an agent guesses Coolify command syntax from stale training data. With it, the agent gets the right mental model (everything is addressed by UUID, you need both a project- and server-UUID to create resources, sensitive fields are hidden by default) plus command-accurate references it loads only when needed — so it stops hallucinating flags and starts shipping.
coolify <noun> <verb> [--flags]
nouns: context · server · project · app · database · service · deploy · private-key · github · teams
coolify-cli/
├── SKILL.md # entry point: mental model, auth, common workflows, command map
└── references/
├── context.md # contexts & authentication (cloud + self-hosted, multi-instance)
├── apps.md # applications: 5 create modes, lifecycle, logs, env, storage, previews
├── databases.md # databases: 8 engines, credentials, backups & S3 retention
├── services.md # one-click service stacks
├── server.md # servers, projects, private keys, GitHub Apps, teams
├── deploy.md # trigger / batch / watch / cancel deployments
└── llms-full.txt # upstream full command catalog (every command + flag)
The agent reads SKILL.md first, then pulls in the one reference file it needs for exact flags — keeping context small.
Personal (all projects on your machine):
git clone https://github.com/itorz7/coolify-cli-skills.git
cp -r coolify-cli-skills/coolify-cli ~/.claude/skills/coolify-cliProject (commit it to a repo, shared with your team):
git clone https://github.com/itorz7/coolify-cli-skills.git
cp -r coolify-cli-skills/coolify-cli .claude/skills/coolify-cliThat's it. Claude Code auto-discovers any SKILL.md under ~/.claude/skills/ or <project>/.claude/skills/. Verify with /skills (or just ask Claude "do you see the coolify-cli skill?").
git clone --depth=1 https://github.com/itorz7/coolify-cli-skills.git /tmp/ccs && \
mkdir -p ~/.claude/skills && cp -r /tmp/ccs/coolify-cli ~/.claude/skills/ && rm -rf /tmp/ccsThe skill is plain Markdown, so any agent that supports the Skills format works:
| Agent | Where to put coolify-cli/ |
|---|---|
| Claude Code | ~/.claude/skills/ or <project>/.claude/skills/ |
| Codex / Copilot CLI | drop the folder where your agent loads skills, or point it at coolify-cli/SKILL.md |
| Anything else | feed coolify-cli/SKILL.md (and references/) as context, or hand it references/llms-full.txt directly |
Tell your agent something like:
"Deploy my-api to Coolify and stream the logs." "Create a postgres database in project X and give me the connection string." "Sync
.env.productioninto thewebapp as build-time vars."
It will reach for the skill automatically. The first thing it should do is confirm which instance it's pointed at:
coolify context list # which instances are configured, which is default
coolify context verify # check the current context's connection + authNeed the CLI itself?
curl -fsSL https://raw.githubusercontent.com/coollabsio/coolify-cli/main/scripts/install.sh | bash # Linux/macOS
brew install coollabsio/coolify-cli/coolify-cli # Homebrew
go install github.com/coollabsio/coolify-cli/coolify@latest # Goirm https://raw.githubusercontent.com/coollabsio/coolify-cli/main/scripts/install.ps1 | iex # WindowsGet an API token from your Coolify dashboard at /security/api-tokens, then:
coolify context set-token cloud <token> # Coolify Cloud
coolify context add -d prod https://coolify.example.com <token> # self-hosted (default)cd coolify-cli-skills && git pull
cp -r coolify-cli ~/.claude/skills/coolify-cli # re-copy over the installed copy- Coolify CLI — https://github.com/coollabsio/coolify-cli
- Upstream
llms-full.txt— https://github.com/coollabsio/coolify-cli/blob/v4.x/llms-full.txt - Coolify — https://coolify.io
This skill is an independent packaging of public Coolify CLI documentation for use with coding agents. Coolify and the Coolify CLI are projects of coollabsio.
MIT