Prompt file imported from davydotcom/omarchy-control4-plugin (
.claude/commands/challenge.md). Fill in{{arguments}}before use. Copyright stays with the author.
Route this challenge to the feature-delivery-lead agent, which manages the
re-investigation flow with the debug-investigator.
Two modes, detected from the feedback language:
| Mode | Trigger language | Behavior |
|---|---|---|
| Layer (default) | "also consider", "I think it's related to", "what about" | Merges the engineer's hypothesis with the existing analysis |
| Reject | "wrong", "not correct", "off base", "re-diagnose", "start over" | Archives current analysis into Investigation History, starts fresh |
Flow:
- Read the existing bug spec at
.hero/planning/bugs/<slug>/spec.md - If no diagnosis exists, error and suggest
/diagnosefirst - Detect mode from the feedback language
- Layer: pass original analysis + engineer feedback to
debug-investigatorwith instructions to produce a merged analysis incorporating both - Reject: archive the current Root Cause / Fix Plan into
## Investigation History, then pass engineer feedback as the starting hypothesis to a freshdebug-investigatorrun - Update the spec with revised findings
- Append a new round to
## Investigation History
The agent must write all findings into the spec file on disk. The spec file is the deliverable, not chat output.
Load the challenge-diagnosis skill for the investigation history format and
re-diagnosis protocol.
Session Title
Set a session title reflecting the challenge (e.g. "challenge: auth-timeout — race condition hypothesis").
Challenge: {{arguments}}
