Monitor

Components

Monitor

Long-running or external processes (e.g. a deployment rolling out, a log file growing, an healthcheck endpoint) stay invisible to an agent until the user explicitly asks about them.

That’s when monitors enter the dance: a shell command the agent starts automatically, with every stdout line it produces delivered back as a notification during the session.

Warning

Monitors are only definable within plugins for now.

Format: Claude Code

~/.claude/plugins/cache/<marketplace>/<plugin>/
└── monitors/
    └── monitors.json

[
    {
        "name": "deploy-status",
        "command": "\"${CLAUDE_PLUGIN_ROOT}\"/scripts/poll-deploy.sh ${user_config.api_endpoint}",
        "description": "Deployment status changes"
    },
    {
        "name": "error-log",
        "command": "tail -F ./logs/error.log",
        "description": "Application error log",
        "when": "on-skill-invoke:debug"
    }
]