Lesson 031: Agent guardrails rehearsal
Focus
Document interfaces between humans, retrieval, and policy engines. Token Agent guardrails rehearsal:31 keeps neighbouring lessons differentiable.
Key ideas
- Thread: Agent guardrails rehearsal · drill v1 · spin
171966. - Habit: pair every model utterance with a trace_id you could paste into Grafana.
- Guardrail: write one RACI bullet referencing this lesson tomorrow.
Deep dive notebook
Synthetic drill artefacts
Agent choreography card
1. Observe transcripts bucket `BUCKET-17`
2. Budget steps `5`
3. Tool whitelist: `retrieve_docs, escalate_human, log_decision`
4. Hard stop triggers: hallucination_budget | escalation keyword `URGENT-3`
Practice
Practice Draft three eval assertions QA must greenlight before launch. — 31 Bump literals mindset by 32.