agentleFS
Sign inStart free
LzdevmendesLzdevmendes/motion-studio/AGENTS.md

motion-studio

AGENTS.md0 starsChanged 29 days agoMITView on GitHub

Instructions for AI coding agents (Claude Code, Antigravity, Codex, Cursor, others) working in this repo. Humans: start at README.md.

  • Reads credentials
  • Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.

    Of the AGENTS.md files we read, the 31,774 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.

  • In this repository7 other files it gives its agents.

The file

We split this AGENTS.md into the rules it gives an agent

6 rules. About 0 are common, 4 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Distinctive:

    Default stack: one HTML file, canvas/SVG, zero dependencies.

    line 11
  2. Distinctive:

    No centered-text-on-gradient with everything fading in.

    line 20
  3. Distinctive:

    Use real product assets and copy when the film is about a product.

    line 21
  4. Very specific:

    Do not commit out/, refs/, .env, model files or MP4s.

    line 23
  5. Distinctive:

    When the skill starts inside a project, it extracts that project's branding and design first and uses it for every slide.

    line 39
  6. Very specific:

    After editing skills/post/, run npm run skill:sync inside packages/post-studio (copies to .claude/skills/post and .agents/skills/post).

    line 40

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “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
AGENTS.md

AGENTS.md

Instructions for AI coding agents (Claude Code, Antigravity, Codex, Cursor, others) working in this repo. Humans: start at README.md.

What this repo is

motion-studio renders motion-design videos from code. A scene is one HTML file with a pure seek(t). render.mjs drives headless Chromium frame by frame and ffmpeg encodes MP4 with sound. The agent skill motion-reel (skills/motion-reel/SKILL.md) is the workflow for making a video here. Load it for any request to make, edit or render a video.

Pipeline

  • Scene = scenes/<name>/index.html exposing window.DURATION, optional window.BPM, and window.seek(t) (paints the exact frame at time t, in seconds).
  • Render: npm run render -- scenes/<name> [--fps 60 --w 1920 --h 1080] -> out/<name>.mp4 + one still per second out/<name>-still-NN.png.
  • Default stack: one HTML file, canvas/SVG, zero dependencies. Use Remotion/HyperFrames only if the user asks.
  • Reference docs: docs/scene-contract.md (API), docs/guide.md (workflow), docs/narration.md (voice).

Rules

  • seek(t) is pure: no setTimeout, requestAnimationFrame, Date.now, Math.random. Same t -> same pixels.
  • Motion = closed-form springs from lib/motion.js (spring, track). No easing curves. Presets: snappy (UI), default (cards/camera), heavy (big type/logos), playful (mascots). Tiny overshoot on UI, none on type.
  • A value with several targets uses track() (one spring per change). Never restart a spring.
  • Cuts and hits land on the beat grid (motion.beat(n, bpm)). Sound: lib/audio.mjs synthesizes on the same timeline; a scene's own audio.wav overrides it.
  • Layout is a function of (w, h), not fixed pixels, so 9:16, 1:1 and 16:9 all render. Reframe type and UI per format, do not crop.
  • No centered-text-on-gradient with everything fading in. Ask for or pick a reference (frame, video, image folder) before designing; extract its palette, type and pacing.
  • Use real product assets and copy when the film is about a product. Never invent UI text or data.
  • Secrets go in .env (git-ignored). Never in prompts, code, logs or commits.
  • Do not commit out/, refs/, .env, model files or MP4s.

Critique loop (mandatory before calling a video done)

  1. Render, read the stills. 2. Score each 1-10: composition, type, motion feel, originality. 3. Write the 3 worst problems. 4. Fix and re-render. Repeat until every score is 8+. Gates for long films: plan -> rig -> stills -> animatic -> full pass -> polish -> audio -> render.

Effort

Medium for small fixes and re-renders, high or above for new films.

Editing this repo itself

  • skills/motion-reel/SKILL.md is the single source of truth for the skill. After editing it run npm run skill:sync (copies to .claude/skills/ and .agents/skills/). CI fails if the copies drift.
  • npm test must pass. Keep code comments and docs in English; Portuguese docs live in README.pt-BR.md and docs/pt-BR/.
  • Scenes keep their own assets inside scenes/<name>/. A scene must render with no network access.

Posts (second skill)

  • The post skill (skills/post/SKILL.md) makes social posts, not videos. Its studio is packages/post-studio (Remotion, its own package.json). Load it for any request for a post, carousel, story, banner or reels.
  • When the skill starts inside a project, it extracts that project's branding and design first and uses it for every slide. Bundled brands are demos only.
  • After editing skills/post/, run npm run skill:sync inside packages/post-studio (copies to .claude/skills/post and .agents/skills/post). CI fails if the copies drift.
  • Remotion has its own license. See THIRD_PARTY.md.
  • The video rules above (pure seek(t), springs, beat grid) apply to scenes/, not to packages/post-studio.

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.

Check your setup, free