FAQ

Is Thymos a chatbot?

No. It is an execution runtime.

Does each interface run its own agent?

No. They can all attach to the same backend run.

What makes it progress through work?

The runtime keeps moving through the task: plan, execute, observe, recover, and continue.

What keeps it controlled?

Writs, policy checks, tool scopes, budgets, approvals, and typed tools.

Can it use local models?

Yes. LM Studio, Ollama, and other OpenAI-compatible endpoints work.