AI agents (MCP) VS Code
Motion Text in VS Code
Add an HTTP MCP server and Copilot's agent mode drafts videos you finish in the editor.
VS Code has first-class MCP support through GitHub Copilot's agent mode. You register Motion Text either by dropping a small .vscode/mcp.json in your workspace or by running the MCP: Add Server command and choosing an HTTP server — VS Code writes the entry for you.
Note VS Code's slightly different shape: servers live under a servers key and each names its transport with type: http. Once it's connected, open the Chat view, switch to Agent mode, and ask for a video; the returned link opens the full Motion Text editor.
Connect VS Code
Add to .vscode/mcp.json, or run MCP: Add Server and choose HTTP:
{
"servers": {
"motion-text": { "type": "http", "url": "https://motiontext.video/mcp" }
}
}Put the JSON in .vscode/mcp.json (workspace) or your user settings, then click Start on the server. Confirm the tools appear in the Chat view's tools picker.
That's the anonymous endpoint — drafting only, no account. For the full toolset on your own projects (and, on Pro, 1080p/4K server renders), use the studio endpoint https://motiontext.video/mcp/studio and sign in when VS Code prompts you. Details on the main MCP page.
Good to know
- MCP tools only run in Copilot Agent mode — not Ask or Edit.
- VS Code uses
serversandtype: http, not themcpServersshape other clients use. - You need an active GitHub Copilot subscription for agent mode.
Questions
Do I need GitHub Copilot?
Yes — MCP tools run through Copilot's agent mode in VS Code. The connector itself is free to add.
Why is the config shape different?
VS Code uses a servers object with an explicit type (http for remote), where Cursor and Windsurf use mcpServers. Same URL, different wrapper.
Can it render MP4 in VS Code?
With the studio endpoint on a Pro account, yes — it queues a server render and returns a download link.