Animation Machine

Animation Machine MCP Server

Give Claude, ChatGPT, Cursor and other AI apps a render button.
Your AI writes an HTML animation; Animation Machine renders it to MP4 or GIF with music, voiceover and captions, in every aspect ratio.

Website · Full docs · Setup · Tools · Example prompts · Guide: videos with Claude

---

Server URL: https://animation-machine.com/mcp Transport: Streamable HTTP (remote, hosted, nothing to install) Auth: OAuth 2.1 (dynamic client registration or client ID metadata documents, PKCE) or a personal API key Pricing: included in the $5/month Animation Machine plan

This repository holds documentation, examples and the MCP Registry manifest. The server itself is hosted by Animation Machine.

Full reference (every tool and parameter, OAuth endpoints, limits, errors): animation-machine.com/docs/mcp · as Markdown for agents: docs/mcp.md · server card: /.well-known/mcp/server-card.json

What it does

AI assistants are great at writing animation code (GSAP, CSS keyframes, Web Animations, Lottie, Three.js, canvas) but can't play it back and record it. This server lets them:

• Render an HTML animation (inline HTML, a multi-file project, or an uploaded zip) to MP4 or GIF, in 16:9, 9:16 and 1:1 from one request
• Finish a video you upload: royalty-free music, AI voiceover, burned-in captions (16 languages), title and end cards, a corner handle, your logo
• Preview a single frame first, so the AI can check its own work
• Use motion-graphics templates (title card, lower third, quote, stat counter) with no code
• Write a social Post Kit (caption, hooks, hashtags) for the finished video

It renders code, frame by frame in a real browser, so text, numbers, colors and timing come out exactly as written. It is not a text-to-video model.

Setup

claude.ai, Claude Desktop, Claude mobile

1. Settings → Connectors → Add custom connector
2. Name: Animation Machine, URL: https://animation-machine.com/mcp
3. Connect, sign in to Animation Machine, Allow

Claude Code

claude mcp add --transport http animation-machine https://animation-machine.com/mcp

Then run /mcp → Authenticate.

ChatGPT

Settings → Apps & Connectors → Advanced → enable Developer mode → Create → URL https://animation-machine.com/mcp, authentication OAuth → sign in and Allow.

Cursor (~/.cursor/mcp.json)

{
  "mcpServers": {
    "animation-machine": {
      "url": "https://animation-machine.com/mcp",
      "headers": { "Authorization": "Bearer am_live_…" }
    }
  }
}

VS Code (.vscode/mcp.json)

{
  "servers": {
    "animation-machine": {
      "type": "http",
      "url": "https://animation-machine.com/mcp",
      "headers": { "Authorization": "Bearer am_live_…" }
    }
  }
}

Anything else

Any client that speaks remote MCP over Streamable HTTP works. stdio-only clients can bridge with:

npx mcp-remote https://animation-machine.com/mcp --header "Authorization: Bearer am_live_…"

API keys: sign in at animation-machine.com → account menu → AI agents (MCP) → Create API key. Keys can be revoked there at any time, and connected OAuth apps can be disconnected.

Tools

Tool	What it does	Hints
render	Render HTML / a files map / an uploaded zip, or finish an uploaded video. Options: aspects, resolution, format, fps, duration_seconds, loop_count, crop_to_fill, music_url, music_volume, voiceover_text, voiceover_voice, captions, caption_style, caption_language, text_overlay, intro_title, outro_title, logo_upload_id. Returns job_ids.	write
render_template	title_card, lower_third, quote or stat from a few fields, plus the same output options.	write
preview_frame	One still frame (image) at at_seconds plus the detected animation length.	read-only
get_job	Status, progress, then download_url (24 h), watch_url (share page), studio_url (open in the editor).	read-only
cancel_job	Cancel a queued or running render.	destructive
create_upload	One-time PUT URL for a local zip / video / logo.	write
search_music	Search 500,000 royalty-free CC0 tracks.	read-only
list_voices	Voices for voiceover_text.	read-only
post_kit	Caption, hooks, hashtags and description for a finished render.	read-only
list_renders	Your 10 most recent renders.	read-only
get_account	Plan, usage, job in progress, limits.	read-only

Typical flow: write the animation → preview_frame → render → poll get_job (it returns retry_after_sec) → share download_url.

Example prompts

• "Make a 5-second animated intro for my YouTube channel 'Build Notes': dark navy, the title slides in letter by letter. Render it as an MP4."
• "Turn this launch announcement into a 10-second animated video and render it 16:9, 9:16 and 1:1 with upbeat royalty-free music."
• "Here are our Q3 numbers. Make a 15-second animated bar chart and render it for LinkedIn (1:1)."
• "I'll upload demo.mp4. Add calm music, captions in Spanish, my @handle in the corner and an end card that says 'Try it free'."
• "Use the stat template counting up to 10,000 with the label 'happy customers', then write a Post Kit for Instagram."

More in examples/.

Limits

Built to stay fast for everyone on a shared render server:

• One job (render or preview) at a time per account; agents queue behind people using the website
• HTML animations ≤ 60 s, GIFs ≤ 15 s, uploaded videos ≤ 3 min (loops included)
• Up to 1080p, 24 or 30 fps, up to 3 aspect ratios per request
• Inline HTML ≤ 2 MB, multi-file projects ≤ 5 MB, uploads ≤ 95 MB
• 200 renders a month (shared with the web app), 5 AI voiceovers and 20 Post Kits a day via agents

Need 4K or longer videos? Every render has a studio_url to keep editing in the browser.

Security & privacy

• OAuth 2.1 with PKCE (S256); tokens are stored hashed and can be revoked from the account menu
• Connected apps can render, read their own renders and search music, nothing else: no billing, no account changes, no other users' data
• Animations render in a sandboxed headless browser behind an egress filter: no access to private networks or cloud metadata
• Download links are HMAC-signed and expire after 24 hours; renders are deleted after 32 days
• Uploaded files and generated media are used only to produce your video. Music comes from Freesound (CC0); AI voiceover and Post Kit use Google Gemini; captions use a self-hosted Whisper model.
• Full policy: animation-machine.com/privacy · Terms: animation-machine.com/terms

Support

[email protected] · animation-machine.com

© Cash Flow Crescendo Inc. Documentation in this repository is MIT-licensed. The hosted service is subject to the Animation Machine terms.