Show & tell: a Claude Code / Agent Skill built on top of coolify-cli

Hi! 👋 First, thanks for building and maintaining coolify-cli — the JSON output (--format=json), the context model, and the overall scriptability made it a great foundation to build on.

I wanted to share something we built on top of the CLI.

What we built

coolify-ops — an Agent Skill (for Claude Code / Codex, installable via skills.sh) that drives coolify-cli with natural language to deploy, operate, and troubleshoot apps / services / databases on a self-hosted Coolify instance.

It's a thin, opinionated layer that just shells out to coolify:

• Translates intent → coolify commands: deploy + follow logs, env sync, lifecycle, backups, troubleshooting, and resource creation (app create / service create / database create).
• Leans on coolify <cmd> --help at runtime instead of hard-coding flags, to avoid version drift.
• Encodes safety rules (confirm destructive ops, never auto--f, never default --is-public for databases).
• Understands both English and Chinese prompts.

npx skills add hifizz/coolify-ops-skill

Built and tested against coolify-cli v1.6.2 / Coolify v4.1.1. No change to the CLI is needed — it's purely a consumer. Sharing in case it's useful to other Coolify users, and happy to take feedback if any patterns are off.

Offer

Happy to contribute usage examples/docs upstream if useful — and if there's a preferred place to list community tooling/skills, point me at it.

(Not a bug report — feel free to relabel/close if Issues isn't the right channel, since Discussions don't appear to be enabled.)

Thanks! 🙏

The issue names no repository files, tests, or entry points. Review the proposed coolify-ops community tooling and its offer to contribute usage examples or docs; done means deciding whether to list the tool, request documentation, or close the non-actionable report.