Introduction Quickstart Authentication MCP ServerNEW API Reference Available Apps

MCP Server — Use Neuroviz From Your AI Assistant

Connect Neuroviz to Claude, ChatGPT, Cursor and other AI assistants — no code required. Connect once, then just ask your assistant to create product photos and videos for you.

What is MCP?

MCP (Model Context Protocol) is an open standard that lets AI assistants use external services as tools. Once you connect Neuroviz to your assistant, it can browse our apps, check prices, submit generations and fetch your results — all by itself, while you talk to it in plain English.

How It Works

Connect Once

Add the Neuroviz MCP server to your assistant with your API key

Just Ask

“Retouch this ring photo for my store” — your assistant picks the right app and settings

Get Results

The assistant waits for the generation and hands you the finished image or video

Server URL

MCP endpoint (streamable HTTP)

https://app.neuroviz.ai/mcp

Uses your regular API key

The MCP server authenticates with the same nv_… keys as the REST API — create one at app.neuroviz.ai/api-portal. Generations cost the same credits as in the app, and only successful runs are charged.

Set Up Your Assistant

Pick your AI assistant and follow the steps. Replace YOUR_API_KEY with your key.

Terminal — one command

claude mcp add --transport http neuroviz https://app.neuroviz.ai/mcp \
  --header "Authorization: Bearer YOUR_API_KEY"

Keep your API key secret!

Anyone with the key can spend your credits. Don't share config files that contain it, and deactivate a key immediately if it leaks — you can create a new one anytime.

Try It

Once connected, talk to your assistant naturally. Some things to try:

Retouch this ring photo for my online store: https://…/ring.jpg

How many NeuroViz credits do I have left?

Put this necklace on a model — here's the product image: https://…/necklace.png

Make a short product video of this bracelet and give me the download link.

What would a 4K retouch of this photo cost before you run it?

Images must be reachable by URL

Your assistant sends image inputs as public HTTPS links (your website, cloud storage, CDN…). If a photo only exists on your device, upload it somewhere first — or ask the assistant to use one of our preset model/background images.

Available Tools

These are the actions your assistant can perform on your behalf:

list_appsBrowse the app catalog with credit prices

get_appFull parameter schema, allowed values and pricing rules for one app

estimate_run_costExact credit cost for a run before submitting it

run_appSubmit a generation (asynchronous — returns a run id)

get_runCheck a run and fetch the result image or video when finished

list_runsRecent run history with statuses and charges

get_balanceCurrent credit balance and plan

list_system_imagesCurated preset gallery (models, hands, backgrounds)

get_usage_guideBuilt-in usage documentation for the assistant

Prefer to integrate with code?

Everything the MCP server does is also available on the REST API.