Connect Crawlens to Claude & other AI tools

Crawlens includes an MCP server. Once it's connected, the AI app you already use can read your crawls and audits, then explain issues, prioritise fixes and write the tickets. It uses your own AI subscription, so there's no API key to set up.

What it does

MCP (Model Context Protocol) is the standard way for AI apps to use outside tools. When your AI app needs Crawlens data, it starts Crawlens's MCP server on your computer. The server reads your saved crawls and audits and hands back compact, paginated results.

• Read-only. It can't start crawls, change settings or delete anything.
• Local. It runs on your computer and opens no network port. Crawlens itself uploads nothing.
• Crawlens doesn't need to be open. The AI app starts the server when it needs it, using Crawlens's own runtime. You don't need to install Node.js.

1. Turn it on in Crawlens

Sharing is off by default. Nothing is visible to AI tools until you switch it on.

1. Open Crawlens and go to Settings → Integrations → AI tools (MCP).
2. Turn on Share audits with AI tools.
3. Under Projects AI tools can read, keep All projects or pick Only these projects.
4. Optional: turn on Hide Search Console and Analytics data to keep clicks, impressions, sessions and URL Inspection results out of what AI tools see.

You need at least one project with an audited crawl. If you haven't crawled anything yet, do that first. The AI has nothing to read until then.

2. Connect your AI tool

The same settings card has a Copy setup for button for each app. Use it: it fills in the exact paths for your installation. The snippets below show what that setup looks like, with placeholder paths.

{
  "mcpServers": {
    "crawlens": {
      "command": "C:\\path\\to\\Crawlens.exe",
      "args": ["C:\\path\\to\\resources\\mcp\\server.mjs", "--user-data", "C:\\Users\\you\\AppData\\Roaming\\Crawlens"],
      "env": { "ELECTRON_RUN_AS_NODE": "1" }
    }
  }
}

claude mcp add crawlens --scope user --env ELECTRON_RUN_AS_NODE=1 -- "C:\path\to\Crawlens.exe" "C:\path\to\resources\mcp\server.mjs" --user-data "C:\Users\you\AppData\Roaming\Crawlens"

[mcp_servers.crawlens]
command = 'C:\path\to\Crawlens.exe'
args = ['C:\path\to\resources\mcp\server.mjs', '--user-data', 'C:\Users\you\AppData\Roaming\Crawlens']

[mcp_servers.crawlens.env]
ELECTRON_RUN_AS_NODE = '1'

{
  "mcpServers": {
    "crawlens": {
      "command": "C:\\path\\to\\Crawlens.exe",
      "args": ["C:\\path\\to\\resources\\mcp\\server.mjs", "--user-data", "C:\\Users\\you\\AppData\\Roaming\\Crawlens"],
      "env": { "ELECTRON_RUN_AS_NODE": "1" }
    }
  }
}

{
  "servers": {
    "crawlens": {
      "type": "stdio",
      "command": "C:\\path\\to\\Crawlens.exe",
      "args": ["C:\\path\\to\\resources\\mcp\\server.mjs", "--user-data", "C:\\Users\\you\\AppData\\Roaming\\Crawlens"],
      "env": { "ELECTRON_RUN_AS_NODE": "1" }
    }
  }
}

Other MCP apps: anything that supports stdio MCP servers works. Use the same command, arguments and ELECTRON_RUN_AS_NODE=1 environment variable as above.

3. Test the connection

Back in Crawlens, click Test connection. Crawlens starts the server exactly the way your AI app will and reports what it found, for example:

If that works but your AI app doesn't show Crawlens, the problem is on the app's side. Restart the app, and see Troubleshooting.

What to ask

Refer to your project by its name or domain. A few prompts that work well:

Tools & prompts

These are what your AI app sees. Results are paginated (50 rows by default) so large sites don't flood the conversation.

Ready-made prompts appear in your AI app's prompt or slash-command menu, where supported:

Privacy & safety

• Off until you turn it on, and you can turn it off at any time. The server then refuses every request.
• Read-only. The server opens your data in read-only mode. It can't crawl, change settings or delete anything.
• No network port. It talks to your AI app over standard input/output on your computer only.
• You choose what's visible: all projects or only some, with or without Search Console and Analytics data.

Crawlens doesn't upload anything, but your AI app sends what it reads to its own AI service as part of the conversation. That follows the AI app's privacy terms, so check them before sharing client data.

Troubleshooting

Crawlens doesn't appear in my AI app

Fully restart the app after adding the setup. Claude Desktop keeps running in the system tray, so quit it from there. For Cursor and VS Code, check the JSON is still valid after pasting: a missing comma breaks the whole file.

“Sharing audits with AI tools is turned off”

Turn on Share audits with AI tools in Settings → Integrations → AI tools (MCP).

“No shared project matches …”

The project isn't in the list AI tools may read, or the name doesn't match. The error lists the projects that are available. Add the project under Projects AI tools can read, or ask using one of those names.

“This project has no audited crawl yet”

Run a crawl of the site in Crawlens first. The AI reads audits, so it needs at least one finished, audited crawl.

“This data was saved by an older Crawlens”

Open the project in Crawlens once. That updates its data, and the AI app can read it again.

It stopped working after I moved or reinstalled Crawlens

The setup contains the path to Crawlens.exe. Copy the setup again from Crawlens (or click Add to Claude Desktop again) so the paths are current. Moving the data folder inside Crawlens is fine: the server follows it automatically.

“… is not valid JSON, so Crawlens left it untouched”

Your claude_desktop_config.json already had a syntax error, so Crawlens didn't change it. Fix or remove the file, then click Add to Claude Desktop again.