crawlwise in Cursor

Point it at the server

{
  "mcpServers": {
    "crawlwise": {
      "url": "https://crawlwise.site/mcp"
    }
  }
}

Restart Cursor, then open Settings → MCP. crawlwise should be listed as connected; authorising it opens a browser for OAuth the first time.

If your organisation blocks OAuth, use a bearer key instead: add "headers": { "Authorization": "Bearer cw_live_…" } beside the url. A key needs a Pro plan or above.

Verify it worked

@crawlwise check the AI crawler access for https://example.com

A per-crawler table is the answer you want. If Cursor reports the server as unavailable, the usual causes are a corporate proxy stripping the /mcp path, or the tool list having been cached from before you authorised — reloading the window fixes the second.

In the editor, on the page you are working on

The MCP tools take a URL, so the useful pattern is to point them at your local development server once it is reachable from the internet — or at the preview deployment. The free tools work on any public URL and need no key, which is why they are the ones worth wiring into an editor agent first.

What it will not do

It reports findings, never a prediction. There is no AI visibility score to fetch because none exists, and a checklist score is our own count of the checks that applied rather than a Google signal. Anything unmeasured comes back as not measured, and a model should not flatten that into a zero.

MCP reference · What crawlwise cannot tell you · Claude Code · Windsurf