MCP Setup
VisionDraft exposes 63+ MCP tools for video editing, generation, and export.
What is MCP?
Model Context Protocol (MCP) lets AI agents call VisionDraft tools directly — upload assets, edit timelines, generate media, and export videos without leaving the conversation.
Cloud MCP
Available to all subscribers. OAuth (Claude, ChatGPT) or API key (Cursor, scripts).
Use OAuth when Claude or ChatGPT asks for authentication — sign in to VisionDraft in the browser. Do not use the old mcp.visiondraft.space link; it is deprecated and will error.
Do not use mcp.visiondraft.space — that URL is deprecated.
Desktop MCP
Runs locally when VisionDraft Desktop is open. Same tools, lower latency for local files.
https://visiondraft.space/api/mcpSupported clients
- Claude — Desktop and Web (OAuth)
- ChatGPT — Connectors (OAuth)
- Gemini — MCP extension
- Cursor — mcp.json configuration
- VS Code — MCP extension
- Any MCP-compatible client — API key auth
Authentication
OAuth (recommended): Sign in through your AI client. VisionDraft handles token exchange automatically. See OAuth guide.
API key: Generate from Settings → API Keys. Pass as Authorization: Bearer vd_... header.
Verify connection
Open MCP Setup in your dashboard. Connection status updates when your agent calls any VisionDraft tool.