Lesson 046: Eval slices that compound
Focus
Swap placeholder nouns for your internal service names immediately. Token Eval slices that compound:46 keeps neighbouring lessons differentiable.
Key ideas
- Thread: Eval slices that compound · drill v6 · spin
233466. - Habit: attach a trace_id to every completion you would paste into an ops dashboard.
- Guardrail: add one RACI bullet for prompt or index changes before tomorrow's standup.
Deep dive notebook
Synthetic drill artefacts
Agent choreography
1. Observe bucket `TENANT-17`
2. Max steps `5`
3. Tools: `retrieve, escalate_human, log_decision`
4. Stop: spend cap | keyword `STOP-4`
Practice
Practice Pair with security on one prompt-injection tabletop scenario. — 46 Bump 28.