How to Make AI Videos with Claude or ChatGPT (an MCP Guide)

Claude and ChatGPT are good at writing a video script. On their own, they can't turn it into a video. With an MCP server (Model Context Protocol), your agent gets real tools it can call. Voxmin's MCP server gives it Voxmin FloStudio, an AI video editor with a timeline, so the agent can finish a video instead of handing you prompts to paste somewhere else.

This guide covers connecting it, the first prompts to try, and how to keep spending under control.

What your agent can do once connected

• Start a film from a prompt. It creates a project and briefs FloStudio's own assistant, which writes the script, plans the cast and shots and renders them with sound.
• Edit the cut exactly. Trims, splits, speed, transitions, text, zooms, colour and effects, several in one undo step. Editing renders nothing, so it costs nothing.
• Make it social-ready. Word-timed animated captions (even behind the speaker), dead air cut out, a vertical reframe that follows the face, and cuts on the music's beat.
• Voice and languages. A voice-over in English, Hindi or another Indian language, or a dub of a finished film into several languages.
• Export. It renders the final MP4 and gives you a download link in the chat.

The same connection also covers Voxmin's pictures, songs, karaoke and face swaps.

Step 1: Add the server to your agent

The endpoint is the same for every client:

https://api.voxmin.com/auth/mcp

• Claude (web, Desktop, mobile): Settings → Connectors → Add custom connector, name it Voxmin, paste the URL. Step-by-step: Connect Claude.
• ChatGPT: turn on Developer mode and add a custom MCP server with OAuth. Step-by-step: Connect ChatGPT.
• Cursor, Claude Code, Codex: add it in their MCP settings; see the MCP overview for each one.

Step 2: Sign in and approve

Your agent opens Voxmin. Log in and approve the app (OAuth 2.1), and you don't need to copy a token. You can remove an app's access later under AI Integrations in your account.

Step 3: Check it works

Ask:

What's my Voxmin balance?

If the agent answers with your credit balance, it's connected.

Step 4: Make something

Try a short one first:

Make a 20-second vertical video of a cat dressed as a pirate on a ship at sunset, with punchy captions. Tell me what it will cost before you render.

Then work on it the way you would with an editor:

Cut the pauses, make the captions go behind the cat, and export it.

Keep spending under control

• Agents are told to state the cost before rendering. FloStudio has an estimate tool, and every price comes from the live price list.
• Tools that post, delete, publish or unlock with credits are marked so the agent asks you first.
• Editing, captions, library sound effects and export cost nothing. Credits pay for generated shots, charged voices and music, and the FloStudio assistant's tokens.
• Voxmin sells credit packs with no subscription, and purchased credits never expire. See pricing.

If your agent shows too many tools

Voxmin has tools for everything it does, and some clients cap how many one server may bring (Cursor allows about 40). Connect only the video tools:

https://api.voxmin.com/auth/mcp?areas=flo_studio

Or list the areas you use, separated by commas. The MCP overview lists them.

Next steps

• See everything the editor does on the AI video editor page.
• Making a story with recurring characters? Read how to keep the same character in every AI video shot.