/wbReview — Command Hub
/wbReview performs a targeted review of specific code changes against a plan. It verifies that every task marked complete was actually completed, catching the single failure mode an audit misses: checkbox optimism. Unlike /wbAudit which evaluates an entire codebase holistically, /wbReview zooms in on a specific diff or plan, making it faster and more focused for day-to-day verification.
🎯 Strategic Position
The default failure mode of an AI worker is marking tasks ✅ without finishing them. /wbReview is the adversarial second pass that asks: "assume the worker was lazy. Where did they cut corners?" It requires a plan file — without one you're at the wrong command.
- After an AI worker finishes a plan — verify every checkbox against code.
- After executing a plan yourself — your own work is what you're least qualified to judge.
- Post-hoc on a hotfix — quick check against the brief plan written under pressure.
- After a docs rewrite — verify nothing was hallucinated.
🛠️ Operating Modes
| Mode | Trigger | Output |
|---|---|---|
| Standard | /wbReview <target> --plan=<plan> | PASS / PASS WITH DEBT / FAIL verdict with task-level detail |
| Act | /wbReview <target> --plan=<plan> --act | Verdict plus automatic fix-up of failed tasks |
| Chain to plan | /wbReview <target> --plan=<plan> --wbPlan | Verdict plus next-step plan generation |
✅ What a useful review contains
- A verdict: 🟢 PASS, 🟡 PASS WITH DEBT, or 🔴 FAIL.
- Task-level detail — which checkbox claims passed but the code didn't.
- A "what the review found that the plan missed" section — side effects, new deps, regressions.
- A score (1–10) with a merge / don't-merge recommendation.
- Re-opened checkboxes in the plan file for failed tasks.
🚫 What it cannot do
| Not this | Use instead |
|---|---|
| Audit the entire codebase | /wbAudit |
| Run tests | /wbTest |
| Fix issues automatically | Combine with --act flag |
| Is this file good? | /wbAudit <file> |
| Pre-release sanity check | /wbAudit |
It refuses if the plan file is missing, malformed, or if the current code state has no detectable relationship to the plan.
📚 Reading Order
- ELI5 — the one-paragraph mental model.
- Practical — step-by-step on a real project.
- Expert — architecture, edge cases, and when NOT to use.
- Examples · Part 1 · Part 2 — annotated transcripts.
- Exhaustive simulation · Live demo.
🔗 Related
wbReview.md— the command reference this hub orients you around./wbAudit— holistic assessment when you don't have a plan./wbPlan— the plan you review against./wbTest— runtime verification after review passes.
Quick Reference
/wbReview <target> --plan=<plan.md> # verify plan execution
/wbReview <target> --plan=<plan.md> --act # verify and autofix
/wbReview <target> --plan=<plan.md> --wbPlan # verify and generate next steps← Home · Commands · Install | wb-flow on npm · flow.wbc-ui.com · wi-bg.com