Plan the route
Name your stops in any order and let the assistant assemble the journey — city to city, trail to trail.
On PyPI · v0.2.0a4 beta
travel-animator ships an MCP server. Connect it to Claude, Cursor or any MCP-capable assistant, describe the trip in plain language, and your AI plans the route, tunes the animation and renders the MP4 — on your own machine.
pip install --pre "travel-animator[mcp]"Overview
The Travel Animator MCP server is a Python package — travel-animator on PyPI — that exposes the Travel Animator route-video pipeline to AI assistants over the Model Context Protocol. Install it, sign in once, and an assistant such as Claude or Cursor can plan a route, style the animation and render an MP4 on your own machine, driven entirely by what you type in chat.
MCP is an open standard for connecting assistants to outside tools. Your client starts the server with travel-animator mcp, discovers the tools it offers, and calls them during the conversation — so there is no separate app to learn and no API keys to wire up beyond your Travel Animator sign-in.
It renders the same journeys the mobile app does, from the same account: the animated routes, the 3D vehicle models and the map styles that Travel Animator is built around. If you would rather build videos by hand, the iOS and Android apps cover the same ground without a terminal.
What your assistant can do
The MCP server exposes the full pipeline as tools, so an assistant can go from a sentence to a finished video without you opening an editor.
Name your stops in any order and let the assistant assemble the journey — city to city, trail to trail.
Adjust how the animation looks and moves — vehicle, route styling, pacing — by asking for changes in plain language.
Export an MP4 on your own hardware. libx264 runs anywhere; pass h264_nvenc for NVIDIA encoding.
Install
About two minutes, assuming Python 3.12 or newer is already on your machine.
mcp extraThe extra adds the MCP SDK, which the base install deliberately leaves out. The package is still in beta, so --pre is required until the first stable release.
pip install --pre "travel-animator[mcp]"Same account as the mobile app. On an interactive terminal this runs a magic-link flow — enter your email, then the 6-digit code. For CI, pass a bearer token instead. Credentials are stored at ~/.config/tada/credentials.json with mode 0600.
travel-animator loginAdd the server to your client's MCP configuration and restart it. Then just ask for a video — the assistant can call auth_status to confirm you're signed in before it starts building a trip.
{
"mcpServers": {
"travel-animator-mcp": {
"command": "travel-animator",
"args": ["mcp"]
}
}
}Try it
No commands to memorise — describe the trip and the look you want the way you would to a person.
Command line
travel-animator CLIThe MCP server is one subcommand of a small console script you can also drive yourself.
travel-animator loginMagic-link sign-in, or store a bearer token for non-interactive use.
travel-animator logoutRemove locally stored credentials. Local only — there is no server-side session revocation yet.
travel-animator render-bundleRender a prepared bundle to MP4 with no network access. Run with --help for the full flag reference.
travel-animator mcpRun the stdio MCP server. Needs the mcp extra.
Before you start
Rendering needs an OpenGL implementation — a GPU driver in production, or Mesa/llvmpipe for CPU-only Linux — plus FFmpeg/PyAV's usual runtime libraries.
Linux, macOS and Windows are all supported. On Windows the GL context comes from WGL, which needs an interactive desktop session — a service can't create one, and there's no software-render fallback.
A limited use licence, not an open-source one: install and run it unmodified to prepare and render Travel Animator content. No redistribution, modification or competing services. Contact Lascade for additional rights.
FAQ
MCP is an open standard for connecting AI assistants to outside tools and data. A client — Claude Desktop, Claude Code, Cursor — launches an MCP server, discovers the tools it offers, and can then call them on your behalf during a conversation. travel-animator ships one such server, so the route-video pipeline becomes something an assistant can operate directly.
Any client that speaks MCP over stdio. That includes Claude Desktop, Claude Code and Cursor — the configuration is the same JSON block in each, pointing at the travel-animator command with the mcp argument.
Yes. The package is a companion to the Travel Animator service, so run travel-animator login first. It uses the same account system as the mobile app, and tools that need a credential fail with a clear message telling you to sign in.
On your own machine. Rendering uses your local OpenGL implementation and FFmpeg libraries, and render-bundle renders a prepared bundle with no network access at all.
The package is still in beta, and pip skips pre-releases by default. Install it with the --pre flag: pip install --pre "travel-animator[mcp]".
No. travel-animator is published under a limited use licence: you may install and run it unmodified to prepare and render Travel Animator content, but not redistribute it, modify it or build a competing service with it. The full terms ship inside the package, and bundled fonts and flag artwork stay under their own OFL and MIT licences.
Yes. The default encoder is libx264, which runs everywhere; pass --encoder h264_nvenc to render-bundle for NVIDIA hardware encoding.
Travel Animator is a mobile app first — build the same animated routes by hand on iOS and Android, no terminal required.
Get the app