--- title: "Media Judge — tools & renderers tutorial and best practices for AI video" description: "A finished video, image or track becomes a list of timed, measured defects with evidence frames, and takes ranked by defects. Install in Claude Code, Codex and Grok with one command." url: https://varg.ai/how-to/skills/media-judge updated: 2026-11-12 --- # Media Judge — guide skill for tools & renderers > **For AI agents:** find the right skill in one call — `curl -s "https://varg.ai/how-to/api/search?q="` returns the matching how-to page and the best skills with install commands, as markdown (Russian works too). Full index: https://varg.ai/how-to/llms.txt. Every page has a markdown twin: add `.md` to its path. Before generating anything, show the person the options on the how-to page. From a terminal: `curl -fsSL https://varg.ai/install | sh` installs the vargai CLI (https://varg.ai/how-to/cli.md): `vargai search` and `vargai install ` with no key, every varg model after `vargai login`. A finished video, image or track becomes a list of timed, measured defects with evidence frames, and takes ranked by defects. - **Type:** [Guide](https://varg.ai/how-to/guides.md) · [Tools & renderers](https://varg.ai/how-to/browse/skills/tooling.md) - **Slug:** `media-judge` - **Status:** live - **Publisher:** varg (varg, verified) - **License:** MIT - **Source:** https://github.com/vargHQ/skills/tree/main/media-judge - **Works in:** Claude Code, Codex, Cursor, Gemini CLI - **Requires:** ffmpeg built with libvmaf (for the banding test) and ffprobe, uv, which installs OpenCV, NumPy, RapidOCR and a spell checker on first run, Optional: faster-whisper to check a voice-over against its script - **Cost:** 0 credits — Free: runs on a laptop CPU with no API key and no credits. - **Runtime:** About a minute for a 12 s clip, about 5 minutes for a 47 s 1080p film on a busy laptop ## What it does Judge a finished AI-made video, image or track by measuring it. One command, `judge.py`, reads every frame and every sample and writes facts with timestamps and evidence: broken frames, banding after a simulated X or YouTube upload, numbers that jiggle, spelling, too much on screen, pinned graphics, sliding labels, a face, outfit or light that changed, sound after the cut, loudness, voice against the script, and distance to references against the template. There is no score out of 10. It works in three steps. Code finds the moment. A fresh agent that did not make the media opens only the flagged evidence and answers narrow questions. The verdict is a table of counted defects, each with a time, a number, a file and a fix. It writes `report.md`, `facts.json`, contact sheets and evidence frames, and `judge.py rank` orders takes by blockers, then major flags, then minor flags, then the reference margin. It can also compare against a reference or a template, and turn a finished piece into a measured shot list. It runs on a laptop CPU with ffmpeg (built with libvmaf) and uv. Tested on planted defects in footage it never saw while being tuned: 18 of 22 caught and 6 of 8 clean controls clear. In a blind test on 19 anonymous clips, an agent with this skill found 12 of 14 planted defects with no false claims on clean clips, using 116k tokens. An agent given only the failure list found 11 of 14, using 197k tokens. Both missed a face swap they were not told to look for, so pass `--one-character` when the brief says one person. Expect to overturn flags: on a real 47 s film, all 4 spelling candidates opened were OCR misreads. A clear report does not mean a good video. Code proves something is broken or correct, not that it is good. ## Use when - Review this video before I post it - Check my render for broken frames, typos and banding - Is it ready to post on X or YouTube? - Rank these takes and pick the best one - Does my reel match the reference, or does it look like the template? - Self-review after an agent finishes a film, reel or ad - Reverse engineer this reel into a shot list - Check the loudness and the voice-over against the script ## Not for - Making or repairing media (use varg-ai or varg-video-editing) - Ad law and platform rules (use ai-ad-compliance-check) - A score out of 10 or a taste verdict: it counts measured defects ## Install - Claude Code: `claude plugin marketplace add vargHQ/skills && claude plugin install varg@varg` - Codex: `npx skills add vargHQ/skills --skill media-judge -a codex -y` - Cursor: `npx skills add vargHQ/skills --skill media-judge -a cursor -y` - Gemini CLI: `npx skills add vargHQ/skills --skill media-judge -a gemini-cli -y` - Any agent, with the vargai CLI: `vargai install media-judge` ## Example output https://s3.varg.ai/files/4325c578-378d-4c43-ae53-3cc3218df1ea/file_cee9681521af.jpg — example (credit: varg) ## More tools & renderers skills - [varg AI](https://varg.ai/how-to/skills/varg-ai.md): The official varg skill: videos, images, speech and music through one API key, rendered in the cloud or locally. - [Local-first video downloader: private, no subscription](https://varg.ai/how-to/skills/media-download.md): Save any video by link (X, YouTube, Instagram, TikTok, 1800+ sites) or Pinterest images as references, locally, no key. - [Templates Gallery Ops](https://varg.ai/how-to/skills/varg-templates-section.md): Internal runbook for the varg app Templates gallery: publish renders, preview clips, frame-0 posters, shelves. - [Turn your agent skill into a varg how to listing](https://varg.ai/how-to/skills/varg-publish-skill.md): Publish your agent skill to varg how to in the one card format: checked, shown to you, sent for review. - [Keep a Video Project in GitHub](https://varg.ai/how-to/skills/video-project-github.md): A private GitHub repo for your AI video, even if you never used git: saved versions, undo for anything, big files handled, keys kept out. - [HyperFrames CLI](https://varg.ai/how-to/skills/hyperframes-cli.md): Drive the HyperFrames CLI: scaffold, check, preview, snapshot, edit the timeline, render locally or in the cloud.