CLI

docksy is a fast, scriptable terminal client for everything in the dashboard, including a live, top-like metrics view.

Install

$ npm install -g @crovenai/docksy
$ docksy login https://docksy.example.com --token dsk_...

Requires Node 20+. The config is stored in ~/.config/docksy/config.json. The environment variables DOCKSY_URL and DOCKSY_TOKEN override it, which is handy in CI.

Commands

Command	What it does
`docksy login <url> [--token]`	Save the instance URL and API token
`docksy status`	Fleet overview with colored usage bars and sparklines
`docksy servers [ls\|add\|rm\|provision]`	Manage servers
`docksy apps [ls\|create\|rm]`	Manage apps
`docksy deploy <app> [--follow]`	Deploy, optionally streaming the build log
`docksy logs <app> [-f]`	Runtime logs, optionally followed
`docksy rollback <app> [deploymentId]`	Roll back to the previous (or a given) good deployment
`docksy env <app> [ls\|set K=V…\|unset K]`	Environment variables
`docksy metrics <server> [--watch]`	A live, top-like TUI with sparklines
`docksy diagnose [server\|app]`	Run diagnose and print the findings
`docksy alerts`	Firing and recent alerts
`docksy open`	Open the dashboard in your browser
`docksy mcp`	Run the Docksy MCP server in-process (stdio)

Every read command accepts --json for scripting:

$ docksy status --json | jq .

Deploy from CI

# .github/workflows/deploy.yml (excerpt)
- run: npx -y @crovenai/docksy deploy api --follow
  env:
    DOCKSY_URL: ${{ secrets.DOCKSY_URL }}
    DOCKSY_TOKEN: ${{ secrets.DOCKSY_TOKEN }}

Most people use the GitHub webhook instead. Deploying from CI is useful when you want tests to gate the deploy.

Edit this page on GitHub