Module 10 — Observability & Debugging
Observability & Debugging
A harness you can't observe is a harness you can't improve.
45
minutes
6
artifacts
Most debugging stops at 'the model made a bad decision' (Q2). The real cause is usually in what the model saw (Q1), what the tool returned (Q3), or what was in context (Q5).
After This Module
01
Implement the 6 observability layers (logs, traces, token accounting, replay, metrics, session diffing).
02
Apply the 5 post-mortem questions to debug a harness failure.
03
Use replay-driven debugging to reproduce failures deterministically.
04
Emit the 8-field per-turn payload and export it through OpenTelemetry into a backend you can actually query.
05
Diagnose "invisible-state" bugs that only surface in sessions longer than 30 turns.
Artifacts