MCP

Use WAMA with AI (MCP)

WAMA can talk to popular AI tools through MCP (Model Context Protocol). After a short setup, you can ask the assistant about products, stock, and sales for one point of sale — the one you choose when you connect.

Server address (same for every tool):

https://mcp.wama.cloud/mcp

Two ways to connect

Sign in with WAMA

1. Add the server address above in your AI tool.
2. When asked for a client id, enter: wama-mcp
3. Complete login in the browser, pick a point of sale, and choose read or read & write.
4. Return to the AI tool — you are connected.

You can revoke access later in the web app: Settings → Integrations.

Paste an API key

1. In WAMA web: Settings → Integrations → Create API key.
2. Choose the point of sale and scope, then copy the key (shown only once).
3. In the AI tool, add the same server address and put the key in the Authorization header as:

Bearer wama_…your_key…

More detail: Integrations and the API guide.

Cursor

1. Open Cursor Settings → MCP (or edit your MCP config file).
2. Add a server named wama with URL https://mcp.wama.cloud/mcp.
3. Prefer sign-in: set client id wama-mcp and scope read (or read_write if needed).
4. Approve the browser login, then try a simple question like “List a few products”.

Example config (sign-in):

{
  "mcpServers": {
    "wama": {
      "url": "https://mcp.wama.cloud/mcp",
      "auth": {
        "CLIENT_ID": "wama-mcp",
        "scopes": ["read"]
      }
    }
  }
}

Paste-key alternative: use the same URL and an Authorization header with Bearer wama_….

Claude Code

In a terminal:

claude mcp add --transport http --client-id wama-mcp wama https://mcp.wama.cloud/mcp

Then run claude mcp login wama (or use /mcp in Claude Code) and finish the browser login.

Claude.ai

1. Open Claude.ai settings for connectors / custom MCP.
2. Add server URL https://mcp.wama.cloud/mcp.
3. If asked for a client id, use wama-mcp.
4. Complete the WAMA browser login and consent.

ChatGPT

1. Turn on Developer mode (or custom MCP / connectors, depending on your ChatGPT plan).
2. Add a custom MCP server with URL https://mcp.wama.cloud/mcp.
3. Prefer client id wama-mcp (no secret) when the UI allows it.
4. If ChatGPT only offers automatic client registration and it fails, use paste an API key instead.

VS Code / GitHub Copilot

1. Open MCP / Copilot MCP settings (JSON).
2. Add an HTTP server pointing to https://mcp.wama.cloud/mcp with OAuth client id wama-mcp.
3. Complete the browser login when prompted.

Example:

{
  "servers": {
    "wama": {
      "type": "http",
      "url": "https://mcp.wama.cloud/mcp",
      "oauth": {
        "clientId": "wama-mcp"
      }
    }
  }
}

OpenAI Codex CLI

Add an HTTP MCP server with URL https://mcp.wama.cloud/mcp and OAuth client id wama-mcp. Follow the current Codex docs for the exact config file. Complete browser login when asked.

Windsurf

Add a remote MCP server https://mcp.wama.cloud/mcp with client id wama-mcp, then finish the browser login.

Other AI tools

If your tool supports remote MCP over HTTPS:

1. Use URL https://mcp.wama.cloud/mcp.
2. Client id wama-mcp (no secret) when using sign-in.
3. Or paste an API key as Authorization: Bearer wama_….

Desktop apps that open a local callback URL (http://localhost… or http://127.0.0.1…) are supported automatically.

What you can ask

With a connected assistant (examples):

• “Search products containing …”
• “What is the stock for product …?”
• “Give me a short sales summary for this week”

Write actions (adjust stock, create or complete a sale) need a read & write key or consent scope. Prefer read until you need those.

Safety tips

• One connection = one point of sale. Create another key or reconnect if you need a different warehouse/store.
• Treat API keys like passwords. Do not share them in chat, email, or public tickets.
• Revoke unused keys under Settings → Integrations.
• For developers building on the REST API (not MCP), see the API guide.