coolify-manager

📁 felixallistar/coolify-manager 📅 10 days ago
4
总安装量
4
周安装量
#53109
全站排名
安装命令
npx skills add https://github.com/felixallistar/coolify-manager --skill coolify-manager

Agent 安装分布

amp 4
gemini-cli 4
antigravity 4
github-copilot 4
codex 4
cursor 4

Skill 文档

Coolify Infrastructure Manager

This skill allows you to manage a self-hosted Coolify instance. It handles deployment lifecycles, resource configuration, and advanced debugging.

🧠 Decision Tree

Step 1: Analyze the User Request. Determine which of the following categories the request falls into and follow the instructions.

A. “Fix it” / “Debug it” (Operations)

User asks: “Why did the build fail?”, “Show me logs”, “Restart the service”

  1. Check Broad Health: Run coolify resources list to see if dependencies (DBs, Redis) are healthy.
    • If multiple resources are unhealthy or exited, check the host server disk space.
  2. Check Status: Run coolify app get <uuid> to see if it’s running or exited.
  3. Get Logs:
    • If build failed: coolify app deployments logs <app_uuid>.
    • If runtime error: coolify app logs <uuid>.
  4. Action: If stuck, coolify app restart <uuid>.

B. “Change it” (Configuration)

User asks: “Update environment variables”, “Change the domain”, “Add a database”

  1. Consult Reference: Read resources/common-workflows.md for specific command patterns.
  2. Environment Vars: Remember that coolify app env sync is safer than adding one by one if the user provides a list.

C. “Delete it” (Destructive Operations)

User asks: “Delete the project”, “Remove this database”

  1. Check CLI Support: Some delete operations (like coolify project delete) are not supported by the CLI.
  2. Use Raw API: Consult resources/api-reference.md for the curl command.
  3. Get Credentials: Run coolify context get <context_name> --show-sensitive to get the base URL and token.
  4. Execute: Make the API call with curl.

⚡ Best Practices

  1. Safety First:

    • Never use --force on delete commands unless the user explicitly requested it.
    • Always verify the context (coolify context verify) before running destructive commands to ensure you aren’t in prod when you think you are in staging.
  2. Resource Discovery:

    • If you don’t know the UUID, search by listing resources: coolify resources list.
    • Always prefer UUIDs over names for critical operations to avoid ambiguity.
  3. Handling CLI Gaps (Raw API):

    • If the CLI lacks a feature, consult resources/api-reference.md for the raw API endpoint.
    • Get context credentials with: coolify context get <name> --show-sensitive
    • Use curl to call the API directly.

📚 Resources

File Purpose
resources/cli-reference.md Complete CLI command reference
resources/api-reference.md Raw API endpoints for operations the CLI doesn’t support
resources/common-workflows.md Recipes for common tasks (env sync, domain changes, etc.)
guides/container_ssh.md How to SSH into containers manually

⚠️ Known CLI Limitations

The Coolify CLI (as of v1.4.0) does not support every API operation:

Operation CLI Support Alternative
Delete project ❌ No Use raw API: DELETE /api/v1/projects/{uuid}
Execute command in container ❌ No SSH to host + docker exec
Force delete database ❌ No Use raw API or Web UI

When using the raw API, always get credentials from the context:

coolify context get <context_name> --show-sensitive