Harness Lab · 09
Failure studio
Break it on purpose: doctor, logs, stuck approvals, bad tool loops — then fix.
~30 min
Requires
Open Harness Part I (Hermes chats; gateway optional)
Assumes
Desktop or CLI install works on a good day; you have not turned YOLO on for the course.
If not ready: Harness 03 — Install →
01
Break auth
Doctor + logs
02
Stuck approval
Timeout path
03
Tool thrash
Stop + scope
04
failure-log.md
Three rows
Why this lab exists
Mastery teaches the happy path. Production is the unhappy path: wrong key, hung approval, agent looping tools, gateway silent. This lab forces three failures so doctor and logs become muscle memory — not panic search.
Safety rules
- ·Do this on a non-production profile if you have one.
- ·Never paste API keys or bot tokens into failure-log.md.
- ·Do not open WAN ports or disable allowlist “to reproduce.”
- ·Stop if you are about to run destructive shell on real data — use a throwaway folder.
Goal is observability, not bricking the host. If chat dies completely, reinstall path is mastery 03 — not this lab.
Failure A — provider / auth smoke
Steps · tap to check0/4
Failure B — approval stuck
Steps · tap to check0/4
Failure C — tool thrash / bad path
Steps · tap to check0/4
failure-log.md template
# failure-log.md
Date / profile: [...]
A Auth: error seen · doctor said · fix
B Approval: mode · timeout · fix
C Tool thrash: prompt · stop method · fix
Secrets: none pasted
Outcome
failure-log.md with three forced failures, what you saw (logs/UI), and the fix for each. No secrets pasted.
Proof · this device0/1
Stored on this device only — not sent anywhere.