Manic MCP Server

Give any MCP-capable AI — Claude Code, Claude Desktop, or a local model — the ability to write, validate, and render Manic animations.

The connected model is the author: this server deliberately has no AI- generation tool and spends no Manic AI credits. Instead it hands the model the complete Manic authoring guide, then lets it validate its own .manic source and render real videos through the Manic platform with your API key.

Tools

The intended loop: guide → write → check → (fix → check…) → render → poll → video — with every generated file optionally saved as a real project document.

Related MCP server: Cognitive Core

Setup

Create an API key at app.maniclang.com/account with scopes: check, render:create, jobs:read, projects:read, projects:write. (Without a key, only manic_authoring_guide works.)

No install needed — npx runs the published package directly.

Claude Code

claude mcp add manic -e MANIC_API_KEY=mk_live_… -- npx -y @maniclang/mcp-server

Claude Desktop / other MCP clients

{
  "mcpServers": {
    "manic": {
      "command": "npx",
      "args": ["-y", "@maniclang/mcp-server"],
      "env": { "MANIC_API_KEY": "mk_live_…" }
    }
  }
}

From source (development)

npm install && npm run build

Then use node /path/to/manic-mcp-server/dist/index.js as the command instead of npx -y @maniclang/mcp-server in any config above.

Other clients

Cursor, Windsurf, Cline, Continue, Zed, LM Studio, and any generic stdio MCP client — see docs/USAGE.md for copy-paste configs, the intended authoring loop, example prompts, the cost model, and troubleshooting.

Configuration

Try it

Ask your model:

"Read the Manic authoring guide, then create a 15-second animation of a pendulum tracing its path, validate it, render it as mp4, and give me the video link."

Links

• Manic · Manic App · API reference

• Manic Create — browser playground

• Manic Workbench — desktop client

• Manic Animate — browser extension