Skip to content
Back to skills

Ai Media Quality Review

ASecurity

Inspect generated images, video and audio before delivery using deterministic metadata, contact-sheet, black-frame and silence checks plus visual and listening review. Use to accept, reject or repair Magic Hour media outputs; it does not generate media or predict audience performance.

  • 1,531 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added November 7, 2026
ai-agentspythongoperformance

Works with

  • cli

Security analysis

A100/100

Pro scans all 4 files and shows the line behind each finding

Scanned November 7, 2026

npx -y skills add hashgraph-online/awesome-codex-plugins --skill ai-media-quality-review --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Ai Media Quality Review?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Ai Media Quality Review
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/hashgraph-online-ai-media-quality-review/badge)](https://www.skillsdirectory.com/skills/hashgraph-online-ai-media-quality-review)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: ai-media-quality-review
description: Inspect generated images, video and audio before delivery using deterministic metadata, contact-sheet, black-frame and silence checks plus visual and listening review. Use to accept, reject or repair Magic Hour media outputs; it does not generate media or predict audience performance.
license: MIT
metadata:
  author: magichourhq
  version: "1.1.0"
---

# AI media quality review

Review the actual downloaded file against its brief and source media. Resolve `scripts/inspect_media.py` relative to this installed skill's `SKILL.md`, not the user's project directory. Run that absolute script path when FFmpeg and FFprobe are available:

```sh
python3 /absolute/path/to/ai-media-quality-review/scripts/inspect_media.py /absolute/path/to/output.mp4 --output-dir /absolute/path/to/review
```

The script creates `probe.json`, `review.json` and a contact sheet, image preview or waveform. For video it also saves `first-frame.jpg` and `last-frame.jpg` from the first and final decoded frames; the contact sheet alone need not include those moments. It reports technical signals; it does not decide whether the creative result is good. A failed rerun clears an earlier `review.json` in a writable output directory before input and dependency checks. Other artifacts may remain from earlier runs; use only the artifacts listed in a newly successful report. Treat a nonzero exit or missing report as unverified, never as a clean result.

## Apply the acceptance criteria

Read the original request and list the few required outcomes and protected details. Inspect the generated artifacts at full size and the intended delivery size.

- **Image:** composition, crop, identity, anatomy, hands, geometry, product/character details, required text and logos, unwanted text and visible artifacts.
- **Video:** first and last frames, intended action and camera, identity/object continuity, flicker, warping, abrupt cuts, actual dimensions, frame rate, duration and audio streams.
- **Speech/music:** listen through the full export; check intelligibility, pronunciation, last word or phrase, clipping, silence, mouth timing, cut points and synchronization.

Inspect `first-frame.jpg` and `last-frame.jpg` as well as the contact sheet. A gradual fade can darken a face without crossing the black-frame threshold; an empty `black_segments` list does not rule it out. Review contact-sheet flags in context. Intentional black or silent sections are not defects. A clean diagnostic report cannot prove smooth motion, accurate lip sync, good music or a compelling result. Full-speed viewing and listening remain required for those claims.

Return a compact verdict for each requirement: pass, fail or unverified, with the observed evidence. A successful request, valid file or plausible thumbnail is not a quality pass.

When repair is authorized, identify the smallest failed requirement and return to the last accepted source. Change the relevant reference, prompt, route, interval or deterministic finishing step. Do not regenerate an accepted shot to fix captions, crop or assembly, and do not continue spending beyond the existing limit.

Deliver the accepted file and retain the review folder with it. State any unavailable playback, listening, source comparison or broad-subject validation.

Files in this skill

  • LICENSE1.1 KB
  • SKILL.md3.2 KB
  • agents/openai.yaml272 B
  • scripts/inspect_media.py7 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.