Troubleshooting

Start with the command that matches your problem:

Command	Use it when
hearthwork status	You want the current state: host, model, context, slots, queue, route, profile, budget.
hearthwork logs [--server] [-f]	You want the host’s log (requests, queue, start and stop, repairs) or llama-server’s own output.
hearthwork doctor	An agent misbehaves and you want to know which layer is at fault.
hearthwork diagnose	Sessions were slow or failed and you want the measured reason.

Pages in this section

• Doctor: a compatibility check with one verdict per agent.
• Diagnose: what limited your sessions, from measured numbers.
• Logs: the host log and the model server’s output.
• Common errors: the exact messages, with causes and fixes.

A good first sequence

1. hearthwork status to see whether the host and model are running and what state they are in.
2. hearthwork doctor --quick to find the layer at fault.
3. hearthwork logs (or logs --server) to read what happened.
4. Still stuck? Open an issue on GitHub with the output of hearthwork doctor and, if you like, hearthwork diagnose --export report.json (numbers only, no prompts or code).