# Review

> (ePost) Use when user says "review", "check my code", "is this good", "look at this before I commit", or "suggest improvements" — detects review type (code quality, a11y, or general improvements) and runs the appropriate review

- Skill: `majiayu000/review-48` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds add majiayu000/review-48`
- Raw SKILL.md: https://api.skillmd.com/api/skills/majiayu000/review-48/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Web & Frontend
- Author: majiayu000 (https://skillmd.com/u/majiayu000)
- Updated: 2026-09-09
- Page: https://skillmd.com/skills/majiayu000/review-48

---


# Review — Unified Review Command

Auto-detect and execute the appropriate review workflow.

## Step 0 — Flag Override

If `$ARGUMENTS` starts with `--code`: load `references/code.md` and execute.
If `$ARGUMENTS` starts with `--a11y`: load `references/a11y.md` and execute.
If `$ARGUMENTS` starts with `--ui`: load `references/ui-mode.md` and execute. Delegate to epost-muji.
If `$ARGUMENTS` starts with `--improvements`: run improvements inline (see below).
Otherwise: continue to Auto-Detection.

## Aspect Files

| File | Purpose |
|------|---------|
| `references/code.md` | Ultrathink edge cases, then parallel verify with reviewers |
| `references/a11y.md` | Review accessibility compliance (WCAG 2.1 AA) |
| `references/ui-mode.md` | Lightweight UI component review by focus area | Loads review/references/ui-mode.md | epost-muji |
| `references/improvements.md` | Review auto-improvement metrics, detect patterns |

## Auto-Detection

Analyze `$ARGUMENTS` keywords:

| Keyword match | Load Reference |
|--------------|----------------|
| "a11y", "accessibility", "wcag" | `references/a11y.md` |
| "ui", "component", "token", "klara", "muji" | `references/ui-mode.md` → epost-muji |
| "improvements", "metrics", "patterns" | Run improvements inline (see below) |
| Default (no keyword match) | `references/code.md` |

## Review-Improvements (Inline)

When dispatching review-improvements, run inline instead of forking (uses haiku model, restricted tools):

1. Read session metrics from `.epost-data/improvements/sessions.jsonl`

2. Read `.epost-data/improvements/sessions.jsonl`
3. Present findings grouped by severity (high → medium → low)
4. For each finding: explain detection, recommended action, next step
5. If no findings: report healthy, show session count
6. Summary table: severity × count

## Execution

For code and a11y reviews: load the reference file and execute its workflow. For improvements: execute inline per the Review-Improvements section above.