What problem does it solve?
The debug skill helps you diagnose why NanoClaw containerized agent runs fail or behave incorrectly, especially when authentication, mounts, environment variables, permissions, or session resumption break.
Core Features & Use Cases
- Container architecture walkthrough: Explains the host-to-container layout, including how environment files, group mounts, IPC, and per-group Claude session directories map into the container filesystem.
- Actionable log and filesystem triage: Points you to the exact host and per-group log files to inspect container stderr/stdout, mounts, and session behavior.
- Common failure fixes: Covers typical breakages such as missing API credentials, root-user/permission mismatches, env vars not reaching the container, incorrect session mount paths, and MCP server startup failures.
- Session persistence validation: Shows how to confirm that Claude session IDs stay consistent for a group and how to clear sessions safely when needed.
Quick Start
Run the debug skill and ask it to diagnose your failure by telling it which group folder you are using and pasting the relevant container log file contents from groups/{folder}/logs/container-*.log.