Review & Fixes
Run reviews and apply blocker fixes. Used by the
/task-review,
/task-human-review, and
/task-review-fix agents.
🤖 Agent-driven
The review and fix agents call these for you. You usually don't run them by hand —
though /task-human-review is exactly where your feedback enters the loop.
Review​
| Command | Purpose |
|---|---|
review-start | Begin a review. Flags: --id, --type ai|human, --by. |
review-end | Finish a review with a verdict. Flags: --id, --verdict approved|fix-needed, --type, --comment. |
Fixes​
| Command | Purpose |
|---|---|
fix-start | Begin fixing review blockers (→ fixing). Flag: --id. |
fix-end | Complete fixes (→ fixed) and record changed files. Flags: --id, --files, --comment. |
Example​
insight-flow review-start --id N42 --type ai --by task-review
insight-flow review-end --id N42 --type ai --verdict fix-needed --comment "2 blockers"
insight-flow fix-start --id N42
insight-flow fix-end --id N42 --files "src/auth.ts" --comment "Fixed blockers 1-2"
A fix-needed verdict routes the task to /task-review-fix; fix-end sends it back
to review. An approved verdict (AI) hands to human review; an approved human
review hands to /task-git. See Transitions & agents.