Claude MCP Integration
ProductAI ships a remote MCP connector for Claude. Add it once, sign in with your ProductAI account, and generate product photos and videos without leaving the conversation.
Overview
The Model Context Protocol (MCP) lets Claude call external tools. The ProductAI connector is a hosted MCP server — there is nothing to install or run locally. Authentication is OAuth: Claude sends you to the ProductAI sign-in you already use, and generations are billed to your account's credits exactly like in the studio.
Features
- Generate product images through natural conversation
- Every ProductAI image model — Nano Banana Pro (default), Nano Banana, Seedream, GPT Image, Kontext
- Product video via Seedance and Gemini Omni
- Upload reference images and reuse them across generations
- Start up to 4 variants in parallel and poll jobs to completion
Add to Claude (web, desktop & mobile)
- In Claude, open Settings → Connectors and choose Add custom connector.
- Name it
ProductAIand paste the connector URL below, then confirm. - Click Connect and sign in with your ProductAI account when Claude opens the authorization page.
https://mcp.productai.photo/mcp
Custom connectors require a Claude plan that supports them. You also need a ProductAI account with credits — generations cost the same as in the studio.
Claude Code (CLI)
claude mcp add --transport http productai https://mcp.productai.photo/mcp
Then run /mcp inside Claude Code to authenticate.
Available Tools
generate_image
Start one or more product-photo generation jobs.
Parameters: image_url, prompt, model, countgenerate_and_wait
Generate image(s) and poll until the results are ready — the one-call version of generate_image + wait_for_job.
Parameters: image_url, prompt, model, countgenerate_video
Start a product video job (text-to-video or reference-to-video).
Parameters: prompt, model, image_url, resolution, duration, aspect_ratioupload_asset
Upload a reference image (base64 or from a URL) and get back a hosted URL to pass as image_url. Max 10MB; PNG, JPG or WebP.
Parameters: data, mime_type, source_url, sha256get_job / wait_for_job
Check a generation job once, or poll it until it completes.
Parameters: job_idExample Usage
Once connected, just describe the shot — Claude picks the tools:
"Shoot this sneaker on a marble table with soft lighting, three options: https://example.com/sneaker.jpg"
Claude uploads the reference, runs the generations in parallel, and returns the finished images in the conversation.