Skip to content

Skills / video-qa

video-qa

Part of the video library (loaded by /video, not run directly): check a rendered video: format, loudness, black/idle/fast/swap, flashing, margins, captions, licences; makes a contact sheet. Run before anything ships.

OfficialRuns scriptsAuditB
Latest
v0.5.2
Aisle
video
Works with
Claude Code
Stars
2

Install

video-qa is a part of the video library. You do not run it yourself: you type /video and describe what you want, and Claude uses this part when it is needed. Installing the library installs this part too.

Install video

npx @agentsaisle/cli add video

Runs scriptsThis skill includes scripts

Your agent may run these scripts on your machine. A maintainer has read every one in full, and the CLI lists the files when you install. You can read them on GitHub first.

Read the files on GitHub

Audit

Checked with aisle audit (version 0.5.2). Run it yourself: aisle audit video-qa

B
Critical
0
High
0
Medium
1
Low
0

9 files · audited 2026-11-03 · rules 1.0.0

MediumAA-P02Can write files and run interpreters without disable-model-invocation

SKILL.md:5

Read Write Glob Bash(node *)

Why The agent may load this skill on its own and then write files and run code in one go.

Fix Set disable-model-invocation: true so the user starts it, or narrow the Bash patterns.

What this skill can use

Tools

  • Read
  • Write
  • Glob
  • Bash(node *)

Scripts

  • scripts/_checks.mjs
  • scripts/_lib.mjs
  • scripts/qa.mjs
  • scripts/run.mjs

Domains mentioned

None

Automated checks found the patterns listed here. An audit reduces risk; it is not a guarantee of safety. How audits work · Read the rules

Show this grade in a README

The badge updates when the registry is rebuilt. Paste this Markdown into the README:

[![aisle audit](https://img.shields.io/endpoint?url=https%3A%2F%2Fwww.agentsaisle.dev%2Fbadges%2Fvideo-qa.json)](https://www.agentsaisle.dev/skills/video-qa)

SKILL.md

Rendered from commit 23948ce

Video QA

Runs the automatic checks on a rendered MP4 and tells you what to fix. The script measures; you read the contact sheet and decide. It changes nothing except qa/report.json and qa/contact-sheet.png. <skill> is the folder that holds this SKILL.md.

Use when

  • A project has a rendered video in out/ and the user wants it checked reaches its QA step.
  • The user asks why a video would be rejected, too quiet, or hard to read on a phone.

Don't use when

  • There is no render yet. Run video-core first.
  • The user wants edits or a fix. Name what failed and the step that fixes it (references/checks.md), and stop.
  • The question is whether the content is true or good. This skill checks the file and the project, not the argument.

Steps

  1. If <project>/lessons.md exists, read it first and follow it; lessons.json changes the limits for this project.
  2. Run node <skill>/scripts/qa.mjs <project> (add --draft for a 540p draft). It needs ffmpeg and ffprobe on the PATH and the video-core skill installed next to this one.
  3. Read the output. Every FAIL blocks publishing; every WARN needs a decision from you or the user.
  4. Open <project>/qa/contact-sheet.png with the Read tool and look at it once: is anything cut off, unreadable or empty? Say what you see.
  5. For each failure give the cause and the fix from references/checks.md. Do not edit the video yourself.
  6. After the fixes, render again and run the script again. Do not publish on a fail unless the user says so in words.

Budget

  • Search first (grep, glob, symbol search). Read line ranges, not whole files.

  • Read at most 8 files per step. If you need more, write down what you learned and narrow the question.

  • Do not re-read a file you just wrote or edited.

  • Prefer the smallest command output that answers the question: head, tail, quiet flags, filtered lists.

  • Run the script once per render. Do not read report.json unless the printed lines were cut at 40.

  • View the contact sheet once, not frame by frame.

Output

  • QA PASS or QA FAIL, each FAIL and WARN in one line with its time, and what you saw in the contact sheet.
  • What was not checked: spelling, platform UI margins and the loudness target are working defaults, not verified rules.

Done when

  • The script exited 0 (or the user accepted each remaining warning), the contact sheet was looked at, and nothing was published, uploaded or installed.

Version history

VersionCommitInstall
0.5.2latest23948ceadd video-qa

Report this skillAgent Aisle on GitHub