chatgpt-sora2-mcp-server

kennethAILAB/chatgpt-sora2-mcp-server

3.2

If you are the rightful owner of chatgpt-sora2-mcp-server and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.

The Model Context Protocol (MCP) server is designed to facilitate communication between language models and various applications, ensuring seamless integration and efficient data exchange.

Sora 2 Conversational Video MCP (No Docker)

An MCP server that lets ChatGPT co-create a video prompt with the user, commit it, and render with Sora 2. The resulting MP4 is returned as a playable card in chat via resource_link.

Features

  • Conversational prompt building: start_session, update_prompt, get_prompt
  • Approval gate: commit_prompt
  • Render: render_from_committed (Sora 2 -> MP4)
  • Long jobs: get_render_status
  • Dev-speed mock adapter (no real Sora needed to test ChatGPT UX)

Quickstart

cp .env.example .env
# Set SORA_ADAPTER=mock for instant testing
npm i
npm run dev

Connect in ChatGPT

  1. Open Settings → Connectors → Developer Mode → Add custom connector
  2. Choose Local command and point to npm run dev
  3. In chat:
    • “Start a session: Lagos Nightlife Reel”
    • “Cinematic, fast cuts, neon reflections, Afrobeats ~110 BPM”
    • “Show the draft” → get_prompt
    • “Lock it” → commit_prompt
    • “Render 1080p” → render_from_committed → Playable MP4 appears

Switch to real Sora

  • Set SORA_ADAPTER=real in .env
  • Provide SORA_BASE_URL, OPENAI_API_KEY
  • Edit src/sora.ts (SoraApiAdapter): replace endpoint paths & body with your official Sora 2 API schema.

Notes

  • Always return resource_link with mimeType: "video/mp4" so ChatGPT renders inline.
  • For very long jobs, poll with get_render_status.
  • Optionally proxy Sora asset_url to your CDN and return your signed URL.

License

MIT