agentleFS
Imaginative-PromptingImaginative-Prompting/claude-video-skills/skills/review-page/SKILL.md

review-page

Skill9 starsChanged 12 days agoMITView on GitHub

Review page, WIP page, post a draft, known issues, feedback, preview loop, waiting review, reviewed, show the creator: use whenever a draft, still set or clip of a video is ready to show, when setting up review for a new project, or when recording the creator's feedback on something posted. Keeps one page where every draft goes up the moment it renders as its own entry, newest first, with its known issues at the top, so all feedback happens in one place and nothing is sent as a bare file path.

4 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

7 rules. 0 are common, 6 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every time something renders that the creator should see: a draft, a still set, a shot clip, a comparison.

    line 10
  2. Distinctive:

    Media is referenced where it lives,: never copied into the page.

    line 25
  3. Distinctive:

    If the project has no page yet, a folder of one JSON file per entry and a static page that lists them newest first and plays their media is …

    line 27
  4. Distinctive:

    Read the stored entry back: after writing it. A refused update can look like success.

    line 60
  5. Distinctive:

    Anything the creator should see is an entry,: never a file path in chat.

    line 61
  6. Very specific:

    When the creator gives notes on an entry, mark it reviewed with their words and the time, and route the notes to the session that owns it, …

    line 67
  7. Distinctive:

    Never mark your own entry as waiting review.

    line 70

“Common” is measured: kept by at least 6 in 10 of the 4 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

The review page

When to use

Every time something renders that the creator should see: a draft, a still set, a shot clip, a comparison. Also when setting up a new project (build the page first), and when they give feedback on something posted. The channel calls its page the WIP page.

What the page is

  • One entry per pass. Each draft, still set or clip is its own entry: a title, when it was added, its known issues at the top, then what changed and what to look at, then the media, then a link to the previous pass of the same thing.
  • Newest first, by when it was added. An edited entry keeps its old place, so every pass the creator should see is a new entry, never an in-place revision of an old one. Keep the earlier passes and their media so the progression can be followed.
  • Review marks are separate from approval. The director marks an entry "waiting review" once it has checked it, and passes at the same thing share a subject so only the newest one waits. When the creator gives feedback, the entry becomes "reviewed", with a one-line note of what they said and when. Reviewed never means approved.
  • Media is referenced where it lives, never copied into the page.

If the project has no page yet, a folder of one JSON file per entry and a static page that lists them newest first and plays their media is enough. One entry might look like this:

{
  "id": "chorus-1-draft-4",
  "title": "Chorus 1, draft 4",
  "createdAt": "2026-10-12T14:05:00-04:00",
  "subject": "chorus-1",
  "knownIssues": ["f1520–1524: the sign's letters flicker (two coplanar faces); fixed in draft 5"],
  "body": "What changed since draft 3, and what to look at.",
  "media": [
    { "type": "video", "src": "renders/chorus-1-d4.mp4" },
    { "type": "image", "src": "renders/chorus-1-d4-loop.avif" }
  ],
  "previous": "chorus-1-draft-3",
  "review": { "state": "waiting" }
}

Posting

  1. Render, scan for blank frames, post at once with the known issues on top, then fix. Never batch several drafts, and never hold one back for a peer's sign-off. The creator wants to watch the work move; a review becomes notes on the posted entry.
  2. Known issues are everything you already know is wrong, in plain words, with frame numbers.
  3. A shot clip carries its own song (muxed in) and plays on its own. A clip that plays against the full song needs its start offset in seconds, never in frames.
  4. Every new version gets a new file name. Players and caches remember a clip by its URL, and the cut may still use the old file. Never overwrite a delivered file.
  5. Looping previews are 10-bit AVIF made from the render's mp4 (ffmpeg -i in.mp4 -c:v libsvtav1 -crf 23 -preset 6 -pix_fmt yuv420p10le -f avif out.avif), never lossy WebP, which broke dark gradients into blocks. Zoom into a flat dark area to check for blocks before posting.
  6. Read the stored entry back after writing it. A refused update can look like success.
  7. Anything the creator should see is an entry, never a file path in chat.
  8. Third-party references stay off the page. Other people's films and footage are for matching locally; the page shows only your own work.

Taking feedback

  • When the creator gives notes on an entry, mark it reviewed with their words and the time, and route the notes to the session that owns it, verbatim (directing-sessions).
  • A session that gets notes directly in its own chat marks its own entry reviewed and tells the director.
  • Never mark your own entry as waiting review; the director does that after checking it.

Verify

  • The new entry is at the top of the page, its media loads, and the clip plays its picture with its sound.
  • Its known issues are the first thing on it.
  • The previous pass is still there, unchanged, and linked.

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.