Connect Your AI Assistant

Shoptera uses a "Bring Your Own LLM" model — you connect your Claude, ChatGPT, or any AI assistant, and Shoptera provides the product data and optimization tools. Your AI runs the agents on your side, using your own account and API credits.

One API key authenticates you, not a single shop. The same key works across every project you own or are a team member of. Your AI calls list_my_projects to discover which shops it can operate on.

Open source on GitHub

All adapter configs, the ChatGPT OpenAPI spec, and example prompts live in github.com/shoptera-ai/shoptera-mcp. Star it, fork it, or open a PR with your own setup recipe.

Getting your API key

Generate a key once and use it across all your integrations and projects.

Open Settings → AI Connector

In your global account settings (not inside a project).

Click Generate API Key

A new key is created and shown to you.

Copy and store securely

The key is shown only once. Paste it into your AI client right away.

Choose your integration

Pick the method that matches your AI tool. All connect to the same Shoptera Merchant MCP server.

Claude Code

Recommended

The fastest path — use the installer prompt from Settings → AI Connector, or register the MCP server manually via CLI:

Terminal

claude mcp add-json shoptera-merchant '{
  "type": "http",
  "url": "https://shoptera.ai/mcp/merchant/",
  "headers": {
    "Authorization": "Bearer YOUR_API_KEY"
  }
}'

Claude.ai (Desktop / Web)

Open Settings → Integrations → Add MCP Server

In the Claude desktop app or claude.ai web interface.

Set server name

shoptera-merchant

Set server URL

https://shoptera.ai/mcp/merchant/

Set auth header

Bearer YOUR_API_KEY

ChatGPT (Custom GPT with Actions)

Create a Custom GPT

Go to chatgpt.com/gpts/create

Import Actions from URL

https://shoptera.ai/api/openapi.json

Set authentication

API Key, header Authorization: Bearer YOUR_API_KEY

Cursor

Add to ~/.cursor/mcp.json (global) or .cursor/mcp.json (per-workspace), then reload Cursor.

~/.cursor/mcp.json

{
  "mcpServers": {
    "shoptera-merchant": {
      "type": "streamable-http",
      "url": "https://shoptera.ai/mcp/merchant/",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Windsurf

Add to ~/.codeium/windsurf/mcp_config.json, then restart Windsurf.

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "shoptera-merchant": {
      "type": "streamable-http",
      "url": "https://shoptera.ai/mcp/merchant/",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

VS Code (Copilot Chat with MCP)

Add to .vscode/mcp.json in your workspace, then enable MCP in Copilot Chat settings.

.vscode/mcp.json

{
  "servers": {
    "shoptera-merchant": {
      "type": "streamable-http",
      "url": "https://shoptera.ai/mcp/merchant/",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

n8n / Make (Integromat)

Use an HTTP Request node pointing to the REST API endpoints. Set the Authorization header with your API key and pass the project parameter on every call. Combine with an AI Agent node (Claude or GPT) that processes the products and sends results back.

More configurations & examples

Pre-built configs, OpenAPI spec, and ready-to-use prompts on GitHub

Example prompts

Once connected, just talk to your AI naturally. Here are some ideas to get started:

“Which of my shops needs the most work?”

“For my-shop, check the feed status and tell me what needs optimization.”

“Run title optimization on all products in baby-centrum.”

“Find missing GTINs for products without barcodes in fantasysperky-cz.”

REST API (any AI / scripts)

Every project-scoped endpoint takes a project query parameter (slug or UUID). Start by listing your shops:

List your shops

curl -H "Authorization: Bearer YOUR_API_KEY" \
  https://shoptera.ai/api/v1/merchant/projects

Check feed status

curl -H "Authorization: Bearer YOUR_API_KEY" \
  "https://shoptera.ai/api/v1/merchant/status?project=my-shop-slug"

Get optimization workflow

curl -H "Authorization: Bearer YOUR_API_KEY" \
  "https://shoptera.ai/api/v1/merchant/workflow?project=my-shop-slug&type=title"

Fetch products

curl -H "Authorization: Bearer YOUR_API_KEY" \
  "https://shoptera.ai/api/v1/merchant/products?project=my-shop-slug&limit=50&fields=id,title,category"

Save optimization results

curl -X POST -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "workflow_type": "title",
    "results": [{
      "product_id": "...",
      "fields": { "title": "..." },
      "confidence": 85,
      "reason": "..."
    }],
    "batch_id": "batch-1"
  }' \
  "https://shoptera.ai/api/v1/merchant/results?project=my-shop-slug"

Key scope & security

The API key identifies you as a Shoptera user. It gives the connected AI read/write access to every project you own or are a team member of, but only through the merchant MCP / REST surface — it cannot change account settings, billing, or team membership.

Rotate the key from Settings → AI Connector if you ever suspect it leaked. The old key becomes invalid immediately.

FAQ

What happens with my data?

Your AI processes the products and sends optimization results back to Shoptera as "suggestions". You review and approve them in the Inbox, just like with Shoptera's built-in agents.

How do I verify it works?

After connecting, ask your AI to "check my feed status". It should return your product count and available optimizations.

Is there a rate limit?

Yes — 60 requests per minute per API key. This is more than enough for typical optimization workflows.

Export Formats

Project Settings