Lesson 059: Eval harness hardening
Focus
Anchor this drill to one production LLM workflow—even hypothetical. Token Eval harness hardening:59 keeps neighbouring lessons differentiable.
Key ideas
- Thread: Eval harness hardening · drill v9 · spin
250393. - 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
Token CFO scratchpad
- prompt_budget: 1802
- completion_budget: 860
- cache_key: `751a`
Hypothesis: halving completions moves P95 ~8% — record actuals.
Practice
Practice Attach rollback steps if cost-per-request crosses your guardrail. — 59 Bump 39.