sentinel-cli

CLI and MCP server for the Sentinel DevOps portal. Deploy, monitor, and manage services from the terminal or via AI agents.

Install

pip install sentinel-cli

Requires Python 3.12+.

CLI

sentinel login                        # OTP auth (Payd Auth)
sentinel projects                     # List all projects
sentinel services                     # List all containers
sentinel status                       # Projects + last deploy overview
sentinel deploy bradar                # Deploy by project name
sentinel deploy bradar --tag v1.2.3   # Deploy specific image tag
sentinel rollback bradar <deploy-id>  # Roll back to previous deploy
sentinel deployments --project bradar # Deployment history
sentinel logs sentinel-api --tail 50  # Container logs
sentinel logs bradar --since 1h       # Logs from last hour

Auth

Run sentinel login once. Tokens are cached at ~/.sentinel/credentials.json with auto-refresh.

Or set SENTINEL_TOKEN env var with a valid admin JWT to skip the login flow.

Override API URL

SENTINEL_URL=http://localhost:8000 sentinel projects

MCP Server (for Claude Code / AI agents)

The package includes an MCP server that exposes Sentinel tools for AI agents.

Add to your Claude Code settings (.claude/settings.json):

{
  "mcpServers": {
    "sentinel": {
      "command": "sentinel-mcp"
    }
  }
}

Available tools

The MCP server reads auth from ~/.sentinel/credentials.json (run sentinel login first) or SENTINEL_TOKEN env var.

What is Sentinel?

Sentinel is a self-hosted DevOps portal for managing Docker container deployments behind Caddy reverse proxy. It provides webhook-based deploys, automatic health checks with rollback, custom domain management with on-demand TLS, and a web UI for monitoring.

sentinel.paydlabs.com | GitHub