Productai

ProductAI Plugin and MCP Server

Codex plugin and MCP server for ProductAI — generate AI-powered product photos.

The repository contains:

• An installable Codex plugin manifest in .codex-plugin/plugin.json
• ProductAI workflow guidance in skills/productai-workflows/
• A self-contained MCP bundle in mcp/productai/
• The TypeScript MCP source in src/

Tools

Tool	Description
generate_image	Generate a product photo from an image URL + prompt
generate_and_wait	Generate and poll until the result is ready (one-step)
get_job	Check generation job status
wait_for_job	Poll until a job completes

Models

nanobanana2.1 (default), nanobananapro (best quality), gpt-low, gpt-medium, gpt-high, gpt-1.5, gpt-2, gpt-2.5, gpt-2.5-precise, kontext-pro, kontext-max, nanobanana, nanobanana2, seedream, seedream-5-lite, seedream-5-pro

• gpt-2.5 — OpenAI GPT Image 2.5, the best GPT model: better images and faster (~11s) than gpt-2. 10 credits.
• gpt-2.5-precise — GPT Image 2.5 precise-edit variant: changes only what the prompt asks and keeps the product, composition and background as they are. Best for small targeted edits to an existing photo. A little slower (~15–20s). 10 credits.
• nanobanana2.1 — Google Nano Banana 2.1, the successor to nanobanana2: better quality at the same 5 credits.

Multi-image input (an array of up to 5 image URLs; up to 10 for seedream-5-lite / seedream-5-pro) is supported with the gpt-1.5, gpt-2, gpt-2.5, gpt-2.5-precise, nanobanana, nanobananapro, nanobanana2, nanobanana2.1, seedream, seedream-5-lite and seedream-5-pro models.

The ProductAI API also accepts gpt-1.5, gpt-2, gpt-2.5 and gpt-2.5-precise without an image_url (text-to-image), but these MCP tools currently always require an image URL.

Setup

Install as a Codex plugin

Install the plugin from this repository, then set PRODUCTAI_API_KEY in the environment used to launch Codex. The plugin starts the bundled MCP server from mcp/productai/index.js.

Get an API key from the ProductAI API access dashboard.

Run the MCP server from source

1. Get your API key

Sign up at ProductAI and get your API key from the dashboard.

2. Build

cd productai-mcp
npm install
npm run build

3. Configure an MCP client

Use the built server in any stdio-compatible MCP client. For Claude Desktop, add this to claude_desktop_config.json:

{
  "mcpServers": {
    "productai": {
      "command": "node",
      "args": ["/FULL/PATH/TO/productai-mcp/dist/index.js"],
      "env": {
        "PRODUCTAI_API_KEY": "your-api-key-here"
      }
    }
  }
}

Tip: If node is not found, use the full path to your Node.js binary in the command field. You can find it by running which node in your terminal (e.g. "/Users/you/.nvm/versions/node/v22.0.0/bin/node").

New to MCP? How to add your first MCP server

If this is your first time adding an MCP server, here's how to get started:

Claude Desktop:

1. Open Claude Desktop
2. Go to Settings > Developer > Edit Config
3. This opens the claude_desktop_config.json file. Paste the config from step 3 above (replace the full path and API key)
4. Save the file and restart Claude Desktop
5. You should see the ProductAI tools available in the chat (look for the hammer icon)

Cursor:

1. Open Cursor Settings (Cmd+, on Mac, Ctrl+, on Windows)
2. Search for "MCP" in settings
3. Click Add new MCP server
4. Add the config from step 3 above
5. Restart Cursor

Claude Code (CLI):

Run this command (replace the path and API key):

claude mcp add productai -- node /FULL/PATH/TO/productai-mcp/dist/index.js

Then set the environment variable:

export PRODUCTAI_API_KEY="your-api-key-here"

Usage Examples

"Generate a product photo of this shoe on a white marble surface" (with image URL)

"Use the kontext-pro model to add sunglasses to this product image"

"Check the status of my ProductAI job 245922"