Sora MCP
Status: This MCP integration is retired.
OpenAI Sora video generation — text-to-video, image-to-video, character references.
Generate short Sora videos from text prompts, animate stills, or carry a character reference across clips. Sora 1 and Sora 2 models supported.
This extension registers the sora MCP server with VS Code so GitHub
Copilot and any other agent that speaks the Model Context Protocol
can call it directly from chat.
Quick Start
- Install this extension. VS Code registers the
sora MCP server automatically.
- Get an API key from Ace Data Cloud (Applications → API Key). New accounts include free trial credit.
- Open Copilot Chat in agent mode and ask for a video task — the extension prompts for the API key the first time and stores it in the OS keychain via VS Code's
SecretStorage.
You can rotate or remove the API key any time from the command palette:
- Sora MCP: Set Ace Data Cloud API Key
- Sora MCP: Clear Ace Data Cloud API Key
The default config talks to the hosted streamable-HTTP endpoint at
https://sora.mcp.acedata.cloud/mcp — no Python, no uvx, no local install needed.
Example prompts
- "Generate a Sora video: time-lapse of a coffee being poured in slow motion. Use sora 2 portrait."
- "Animate https://example.com/dog.jpg into a Sora clip of the dog walking through a meadow."
10 tools available via this server.
| Tool |
Description |
sora_generate_video |
Generate an AI video from a text prompt using Sora. |
sora_generate_video_from_image |
Generate an AI video from reference images using Sora (Image-to-Video). |
sora_generate_video_with_character |
Generate an AI video featuring a character from a reference video. |
sora_generate_video_async |
Generate an AI video asynchronously with callback notification. |
sora_generate_video_v2 |
Generate an AI video using Sora Version 2 (partner channel). |
sora_generate_video_v2_async |
Generate an AI video asynchronously using Sora Version 2 with callback. |
sora_get_task |
Query the status and result of a video generation task. |
sora_get_tasks_batch |
Query multiple video generation tasks at once. |
sora_list_models |
List all available Sora models and their capabilities. |
sora_list_actions |
List all available Sora API actions and corresponding tools. |
Supported Models
sora-1, sora-2
Pricing
From $0.50 per clip. Free trial credit on sign-up.
Configuration
This extension implements the mcpServerDefinitionProviders contribution point
and registers a single hosted server with VS Code:
Provider id : acedatacloud.sora
Server label: Sora MCP
Server URL : https://sora.mcp.acedata.cloud/mcp
Transport : Streamable HTTP
Auth : Bearer API key from VS Code SecretStorage (or $ACEDATACLOUD_API_TOKEN)
You don't need to edit mcp.json — the extension handles registration and
token handling automatically. If you'd rather configure things by hand, the
sections below show equivalent mcp.json snippets you can use instead of
this extension.
Alternative: manual mcp.json (hosted)
{
"servers": {
"sora": {
"type": "http",
"url": "https://sora.mcp.acedata.cloud/mcp",
"headers": { "Authorization": "Bearer ${input:acedatacloud_api_token}" }
}
},
"inputs": [
{
"type": "promptString",
"id": "acedatacloud_api_token",
"description": "Ace Data Cloud API key",
"password": true
}
]
}
Alternative: local stdio (no network roundtrip)
For offline dev, air-gapped environments, or pinning to a specific PyPI
version, install uv and use:
{
"servers": {
"sora": {
"type": "stdio",
"command": "uvx",
"args": ["mcp-sora"],
"env": { "ACEDATACLOUD_API_TOKEN": "${input:acedatacloud_api_token}" }
}
}
}
uvx will download and run the latest mcp-sora on demand.
Links
License
MIT — see LICENSE.
