Provides structured, actionable feedback when rejecting work that failed quality review, and creates child beads to track required fixes.
You are rejecting work that did not pass the inquisitor's quality review. Your feedback must be specific and actionable enough that a grunt — who does not think creatively — can fix the issues without ambiguity.
Review the inquisitor's findings for the bead.
Categorize each issue by severity:
For each critical or major issue, create a child bead:
bd create "<concise fix description>" -p <priority>
bd dep add <parent-bead-id> <child-bead-id>
Add structured feedback as a comment on each child bead:
bd comment <child-bead-id> "<feedback using the format below>"
Block the parent bead until children are resolved:
bd update <parent-bead-id> --status blocked
For each issue, provide:
### Issue: <short descriptive title>
- **Severity**: Critical | Major | Minor
- **Location**: <file path>:<line number(s)>
- **Problem**: <What is wrong. Be specific — reference actual code.>
- **Expected**: <What it should look like or do instead.>
- **Fix**: <Exact steps to resolve. Include code snippets showing the correct implementation if possible.>
Once all child beads are created, re-prompt the grunt using the grunt-prompt skill. The prompt must include:
Track the retry count. If a bead has been rejected 3+ times for the same issue, escalate to the user — something systemic may be wrong.