# QA Reroll

> Score generated Brand Shoot Kit images for product accuracy, realism, brand fit, commerce usefulness, and artifact risk. Use for QA reports, reject queues, reroll decisions, and correction prompts.

- Skill: `themattberman/qa-reroll` (Agent Skill)
- Install (CLI): `npx skillmds@latest add themattberman/qa-reroll`
- Raw SKILL.md: https://api.skillmd.com/api/skills/themattberman/qa-reroll/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: TheMattBerman (https://skillmd.com/u/themattberman)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/themattberman/qa-reroll

---


# QA Reroll

Do not pass pretty failures.

## Score Gates

- product accuracy
- label/logo/text preservation
- realism/artifact risk
- brand fit
- channel usefulness
- composition clarity

## Reject Immediately

- changed packaging shape, count, or label text
- malformed hands or body interaction
- fake badges/certifications/claims
- product obscured or too small
- wrong variant/flavor/material

Use `scripts/qa-images.py --packet <packet>` to write structured QA results. Failed shots should produce concrete reroll instructions, not vague criticism.

Executable module owner path:
- `scripts/modules/qa_reroll.py --packet <packet-dir> --run-qa` -> updates `qa-results.json` and writes `reroll-manifest.json`