📖About & Usage

Owner agent — accountability this droid serves

Guardian is the Director of Infrastructure and must ensure all systems stay healthy and resilient. This droid covers the monitoring slice — continuously watching Carol for problems so Guardian knows what's broken and can act on it.

Droid responsibility

Shield runs a two-stage health check on Carol every 5 minutes: verifying the worker thread is alive and then verifying the Claude CLI can actually respond. It records every check result to the database, giving Guardian the visibility he needs to spot problems early and decide what to do.

What the droid actually does

• Checks Carol's worker thread via /api/stats endpoint every 5 minutes
• Checks Claude CLI responsiveness via /api/probe endpoint (15-second timeout)
• Marks status 'ok' only if both checks pass; 'down' if either fails
• Records timestamp, response time, status, and error details to the database in Berlin timezone

Boundaries

• Only observes and records — doesn't restart services or fix problems
• Doesn't manage tunnels or monitor other infrastructure components
• Limited to Carol health checks; doesn't cover the Guardian apps themselves

👤Owner

📚Recent initiatives

Initiatives that touched this droid — a short summary each; open one for the full story.