Build with AI
Connect Cut.Pro to Claude, ChatGPT, Cursor or VS Code over MCP, and give your coding agent the documentation as plain text.
Cut.Pro has an official MCP server. Once connected, the assistant uses every route of the v1 API as a tool: it clips a video, generates image, video and voice, creates personas, renders and publishes. In clients that support confirmation, it shows the price and waits for a yes before charging.
| Path | Who it is for | Authentication |
|---|---|---|
Local, via npx | Claude Code, Claude Desktop, Cursor, VS Code | API key, on the Gratuito, Básico, Plus, and Pro plans |
Remote, https://mcp.cut.pro | ChatGPT, Claude.ai | OAuth, on any plan |
Local
claude mcp add cutpro \
--env CUTPRO_API_KEY=$CUTPRO_API_KEY \
-- npx -y @cutpro/mcp@latestA key with access to several workspaces also needs CUTPRO_WORKSPACE_ID in env. Generate the key in Authentication.
Remote
Add the connector
In ChatGPT, under Connectors. In Claude.ai, under Settings, Connectors, Add custom connector. The URL is https://mcp.cut.pro.
Authorize
The client opens cut.pro. Sign in, choose the workspace and click Connect. Jobs use that workspace's credits. Disconnect whenever you want in Connected apps.
The remote server follows OAuth 2.1 with metadata discovery, dynamic client registration, PKCE and refresh tokens.
Ask in natural language
Analyze this YouTube video, generate the clips and send me the 3 highest-scoring ones already rendered.
The assistant chains analyze_video, create_submission, get_submission until it finishes, list_clips, render_clip, get_render and download_render.
Make a 5-second vertical video of a barista drawing in the cafe with Veo. Show me the price first.
It reads models and prices with list_generation_models, shows the total and, after the yes, calls create_generation and get_generation. To publish the result, open_generation_edit returns the edit_id.
Each tool has the API operation's name in snake_case, so what is in the reference applies to the assistant. The server also ships ready-made prompts that show up in the client menu: generate_video, product_ad, create_persona, clip_and_schedule and transcribe.
Give the documentation to your coding agent
All the documentation is available as plain text, in the format models read directly:
| Address | What it contains |
|---|---|
/docs/llms.txt | Index of all pages, in the three languages. |
/docs/llms-full.txt | The entire documentation in one file. |
/docs/<page>.md | Any page in Markdown, like /docs/quickstart.md. |
openapi.json | The complete v1 API contract. |
To get the integration written into your project, paste this into Cursor, Claude Code or the assistant you use:
Read https://cut.pro/docs/llms-full.txt and the contract at
https://api.cut.pro/api/v1/openapi.json.
Write a script that:
1. analyzes the video URL I pass in (POST /clips/info) and shows the cost
in credits before spending;
2. creates the submission and follows it until it finishes;
3. renders the highest-scoring clip and returns the MP4 link.
The key comes from the CUTPRO_API_KEY environment variable and goes in the X-Api-Key header.
Every field is snake_case and the resource's own id is always "id". Every error comes
as { code, extra }: handle it by code and respect Retry-After on a 429.The key gives access to the workspace's credits and connected accounts. Keep it in an environment variable and never paste the value into the code or into the request to the assistant.