MCP server

Endpoint

https://wpaq.com/mcp

Header

Authorization: Bearer wpaq_…

Plan

Pro or Business, checked on every request. A downgrade returns 403. The key stays on the account.

REST reference

Tool results match the REST API. Create a key under Dashboard → Integrations.

Add this to Cursor MCP settings (mcp.json). Replace the key with the one shown once at creation.

{
  "mcpServers": {
    "wpaq": {
      "url": "https://wpaq.com/mcp",
      "headers": {
        "Authorization": "Bearer wpaq_..."
      }
    }
  }
}

The same shape goes in Claude Desktop's config (claude_desktop_config.json). Restart Claude after saving.

{
  "mcpServers": {
    "wpaq": {
      "url": "https://wpaq.com/mcp",
      "headers": {
        "Authorization": "Bearer wpaq_..."
      }
    }
  }
}

Project or user .mcp.json uses type: http for a remote Streamable HTTP server.

{
  "mcpServers": {
    "wpaq": {
      "type": "http",
      "url": "https://wpaq.com/mcp",
      "headers": {
        "Authorization": "Bearer wpaq_..."
      }
    }
  }
}

Or register it from the terminal:

claude mcp add --transport http wpaq https://wpaq.com/mcp \
  --header "Authorization: Bearer wpaq_..."

Any client that speaks MCP Streamable HTTP can use the same URL and header. The client sends the protocol framing. You do not hand-write JSON-RPC.

URL: https://wpaq.com/mcp
Authorization: Bearer wpaq_...
Transport: Streamable HTTP

A missing or invalid key is 401 with {"detail":"Invalid API key"}. An account that is not Pro or Business is 403. Tool failures inside a call come back as 401: …, 402: …, 403: …, 404: …, or 429: … followed by the same message the REST API uses. See Errors.

1. Call start_site_audit, start_ai_visibility_audit, start_conversion_audit, or start_quality_audit.
2. Read id and status from the JSON text the tool returns.
3. Call the matching get_* tool until status is completed, completed_with_warnings, or failed. Wait a few seconds between calls.
4. Read findings, the report, or a share link only after the run has finished. A share on an unfinished Site Audit returns 409.

AI Visibility, Conversion, and Content Quality charge 10 credits on a DNS-verified site and 20 otherwise, including third-party hosts. Site Audit uses the monthly page quota, with an unverified cap of 25 pages on Pro and 100 on Business. start_site_audit is limited to 40 starts per hour per account and 8 per hour per IP. The three AI start tools are limited to 300 per hour on a paid plan. Polling is not on those limiters. A 429 means back off.

Arguments below are what you pass. The client also receives a request context. It is not an argument you set. Results are JSON text with the same fields as the matching REST route.

You ask the agent to audit a site. It starts the job, polls, then summarizes. The JSON below is an example.

Run a Site Audit on https://example.com and tell me the score when it finishes.

{
  "url": "https://example.com",
  "crawl_mode": "bfs"
}

{
  "id": "aud_example",
  "url": "https://example.com",
  "status": "queued",
  "scan_type": "full"
}

{ "audit_id": "aud_example" }

{
  "id": "aud_example",
  "status": "completed",
  "scores": { "overall": 86 }
}

The agent then says the example.com Site Audit finished with an overall score of 86. For an AI Visibility, Conversion, or Content Quality run, swap in the matching start and get tools. Call get_conversion_questionnaire first when you want to pass answers.