Command reference

Running nod without arguments opens the interactive menu. All commands below are accessible from this menu — select an option and follow the prompts. This is the recommended way to use the CLI.

nod

Interactive setup wizard. Configures agent credentials, detects project resources, and sets permission rules.

nod connect

The wizard walks through:

• Agent ID and Secret (from the Nod app)
• Working directory selection
• Resource detection (MCP servers, skills, subagents)
• Per-agent resource scoping (which MCP servers, skills, subagents to enable)
• Auto-execute categories (file reads, file edits, shell commands, web access)

Config is saved to ~/.nod/agents/<name>/config.json. A PreToolUse hook and tool scripts are generated for Claude Code.

Start an agent as a background service using the OS service manager (launchd on macOS, systemd on Linux).

nod up my-agent

The agent connects to Nod via WebSocket and begins listening for messages and tasks. It persists across terminal sessions and restarts.

Stop a running background agent.

nod down my-agent

Run an agent in the foreground (this terminal). Useful for debugging.

nod start my-agent

Press Ctrl+C to stop.

List all configured agents with their current status.

nod ps

# Output:
# NAME          STATUS      PID     PORT
# my-agent      running     12345   47112
# deploy-bot    stopped     —       —

View agent logs. Follows by default (like tail -f).

nod logs my-agent

Remove an agent. Stops the service and deletes local configuration.

nod rm my-agent

Show full agent details including resources, security rules, and session info.

nod status my-agent

Interactive configuration menu with three sub-menus:

nod config my-agent

Run diagnostics to verify everything is working.

nod doctor my-agent

Checks:

• Claude Code is installed and accessible
• Agent credentials are valid
• Backend connection is working
• Config files are present and valid
• Hook server port is available

For production, set NOD_API_URL=https://api.asknod.ai.