agentleFS
neskuchnyneskuchny/content_factory/.claude/skills/jury/SKILL.md

jury

Skill1 starsChanged yesterdayMITView on GitHub

Jury: a short creative review of a video draft after the technical QC. Six questions with timecodes (hook, key moment, \"shot like everyone else's\", the weakest 5 seconds, sound at the cuts, generated text), a score of 5–9 and at most three fixes. Use for the reviewer as a second pass on every video that has passed the technical checks, and when the human asks \"did it turn out interesting\".

7 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

1 rule. 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 6 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Sound on the picture: do the three main cuts/hits each have their own sound?

    line 25

“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Jury: is this interesting to watch

The idea and the questions are based on references/jury.md from sarpstar/cinecode (MIT). Adapted for the factory.

The technical QC answers "is there any defect". The jury answers "will people watch to the end and forward it". This is the reviewer's second pass after the technical PASS. It is not a new render and not a lengthy check: 2–3 minutes over the already finished materials.

What to look at

  • a contact sheet with the platform's safe zones: python3 .claude/skills/ffmpeg-skill/scripts/look.py <video> --safe reels --tiles 3x2 -o <png>;
  • at most three individual frames at the moments the questions are about;
  • beats.md / the script and the on-screen text;
  • the output of python3 scripts/qa/words_check.py <file with texts>.

Six questions: a timecode and one sentence for each

  1. Hook: in the first 2 seconds, is the object/character already doing something? If not, what to show instead.
  2. Key moment: where is it, and does it hit harder than everything around it?
  3. "Shot like everyone else's": is there a shot that would fit any video from any company (template, stock, AI default)? Which one?
  4. The weakest 5 seconds: where and why (static, repetitive, unclear, drawn out)?
  5. Sound on the picture: do the three main cuts/hits each have their own sound?
  6. Words: does any line sound generated (per words_check.py and by ear reading the text)?

Then: at most three fixes

The most significant first. Each one: a timecode, what to change and how, in numbers. At least one should raise the "wow", not only polish. Cost of the fix: 0 credits (done in code) or how many credits.

Score (honest, against anchors)

5 — student work · 6 — template · 7 — solid professional · 8 — strong agency · 9 — award shortlist. First drafts are usually 5–7.

Where to write

In review.md, a "Jury" section: six answers, the score, up to three fixes. In the approval pack, one line: the score and the main fix; the human chooses which fixes to apply. The jury does not block release: only the technical QC blocks. A score below 6 is a reason to propose a redo to the human, not an obligation to do one.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.