What lives here
Guides are worked examples with runnable config. Every guide answers a single "how do I…" question end-to-end.
| Guide | Answers |
|---|---|
| Scheduled tasks | How do I make rousseau nag me on a schedule via WhatsApp? |
| Self-hosted vLLM | How do I point rousseau at a vLLM endpoint on my internal network? |
| Kubernetes deployment | How do I run rousseau as a Kubernetes Deployment? |
| Audit + approval policies | How do I lock down the bash tool and still let the agent do useful work? |
| Observability | How do I get rousseau's slog output into Loki / Grafana / Datadog? |
| Read-only mode | How do I run rousseau as a read-only inspection agent that never mutates the workspace? |
When to read guides vs. concepts vs. reference
- Concepts — you want to understand how the agent loop works.
- Guides — you want to solve a concrete operational problem.
- Reference — you want the exact signature of a flag.
Next
- Scheduled tasks — often the first guide operators reach for.