The full upstream README, mirrored here for reference. Install config, tool schemas, adoption signals, and an original overview live on the Idea to Motion listing page.
Make animated videos by asking your AI client. You connect one MCP server, then ask for a video in the same conversation you ask for anything else.
Your assistant writes the script, plans each scene and writes the animation code. This server synthesizes the voiceover, aligns the animation to word-level timestamps, renders the frames and hands back an MP4. There is no timeline to edit and nothing to install.
https://api.ideatomotion.com/mcp (Streamable HTTP)com.ideatomotion/idea-to-motionThis repository is the public install reference. The platform itself is closed-source; everything here is configuration and documentation.
Mint a key at ideatomotion.com/connect. Clients that speak OAuth 2.1 (Claude web and Desktop) need no key at all — you authorize in the browser.
| Client | Auth | Config file |
|---|---|---|
| Claude Code | API key header | ~/.claude.json |
| Claude web & Desktop | OAuth (no key) | — |
| Cursor | API key header | ~/.cursor/mcp.json |
| VS Code (Copilot) | API key header | .vscode/mcp.json |
| Antigravity | API key header | ~/.gemini/antigravity/mcp_config.json |
| Windsurf | API key header | ~/.codeium/windsurf/mcp_config.json |
| Gemini CLI | API key header | ~/.gemini/settings.json or .gemini/settings.json |
| Codex CLI | API key header | ~/.codex/config.toml |
| Kiro | mcp-remote bridge | .kiro/settings/mcp.json |
| Zed | mcp-remote bridge | ~/.config/zed/settings.json |
Full copy-paste setup for each, with screenshots: ideatomotion.com/connect.
Three kinds of video:
explainer — narrated. Your AI writes the script; the platform synthesizes
the voiceover and times the animation to word-level transcript boundaries.motion_graphic — silent, on-screen text set to music. Needs no voice
key, so it is the fastest way to a finished file from a standing start.recording — you upload your own video; your AI picks the cuts and the
platform removes the silences and animates around you.Aspect and length are independent: a 15-second 16:9 clip and a 5-minute 9:16 explainer are both valid.
Narration runs on your own ElevenLabs or Cartesia account, connected at /connect. That is a separate bill from that provider — we take no cut and never see your key after it is stored. Silent motion graphics need no voice key.
Creating and previewing are free and unlimited. You pay only to export a finished MP4 — from $1.00 per minute at 1080p, billed to the second. 720p and below are free with a watermark, with a monthly allowance. Full rate card: ideatomotion.com/pricing.
There is deliberately no export MCP tool. Exporting spends real money at a
quality the user chooses, so it is an action you take on the platform — not one
your agent can take on your behalf.
Config field names differ per client and fail silently when wrong — a mistyped key leaves the server simply absent, with no error:
servers here — not mcpServers like other clients.serverUrl, not url. Restart after saving.httpUrl selects Streamable HTTP; plain url would mean SSE.http_headers and use bearer_token_env_var = "IDEATOMOTION_KEY" instead.mcp-remote bridge carries your key over HTTP.url + an Authorization: Bearer header.mcp-remote is only needed for clients that can spawn stdio processes but not call HTTP.MIT — applies to the contents of this repository (configuration examples and documentation), not to the hosted service.