GeoDaddyGeoDaddy

MCP Server

The GeoDaddy MCP server exposes the CLI as a tool that AI assistants can call directly. Once configured, you can ask Claude or Cursor to analyze any URL and get a full GEO report without leaving your conversation.

What is MCP?

Model Context Protocol (MCP) is an open standard that lets AI assistants call external tools. The GeoDaddy MCP server wraps the CLI binary and registers a single tool — analyze_url — that accepts the same parameters as the CLI.

No API keys. No cloud. The analysis still runs entirely locally.

Installation

The MCP server is published on npm as geodaddy-mcp. On first install it automatically downloads the correct CLI binary for your platform. Node.js 18 or later is required.

Claude Desktop

Add the following to your Claude Desktop config file:

  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
  • Windows: %APPDATA%\Claude\claude_desktop_config.json
{
  "mcpServers": {
    "geodaddy": {
      "command": "npx",
      "args": ["-y", "geodaddy-mcp"]
    }
  }
}

Restart Claude Desktop. You should see a hammer icon in the chat input — that confirms the server is active.

Claude Code

Run this command once to register the server:

claude mcp add geodaddy -- npx -y geodaddy-mcp

Claude Code will pick it up automatically in future sessions.

Cursor

  1. Open Settings → MCP Servers
  2. Click Add
  3. Set the command to npx -y geodaddy-mcp
  4. Save and restart Cursor

From source

If you prefer to build the binary yourself:

# 1. Build the CLI binary
git clone https://github.com/borabiricik/geodaddy-cli.git
cd geodaddy-cli
cargo build --release
 
# 2. Install and build the MCP server
cd mcp
npm install
npm run build

The server falls back to target/release/geodaddy automatically, so no extra configuration is needed.

Using the Tool

Once configured, just describe what you want in natural language:

Analyze https://example.com for GEO issues

Crawl my site https://mysite.com up to 30 pages and show me everything that's failing

Check if any AI bots are blocked on https://example.com

Run a full audit of https://example.com including Core Web Vitals

The AI will call analyze_url with the appropriate parameters and return the results inline.

Tool Reference

analyze_url

Runs the GeoDaddy CLI against a URL and returns either a JSON report or a formatted text report.

ParameterTypeRequiredDescription
urlstringyesThe URL to analyze. Supports http://localhost and http://127.0.0.1 for local dev.
max_pagesnumbernoEnable site-wide crawling, stop after N pages. Without this, only the given URL is analyzed.
enable_jsbooleannoEnable JavaScript rendering. Downloads Chromium (~150 MB) on first use.
vitalsbooleannoMeasure Core Web Vitals (LCP, FCP, CLS, TTFB, TBT). Requires headless browser.
fail_undernumber (0–100)noReturn an error if the overall score is below this threshold.
beautybooleannoReturn a human-readable text report instead of JSON. Defaults to JSON.

Return value

By default the tool returns the raw JSON report. Set beauty: true to get formatted text output — useful when you want the AI to interpret results inline rather than parse JSON.

{
  "schema_version": "1",
  "url": "https://example.com",
  "crawled_at": "2026-03-23T15:42:30.123Z",
  "score": 82.5,
  "categories": {
    "technical": 85.0,
    "content": 80.0,
    "geo": 75.0,
    "performance": null
  },
  "pages": [
    {
      "url": "https://example.com/",
      "robots_blocked": false,
      "score": 82.5,
      "categories": { "technical": 85.0, "content": 80.0, "geo": 75.0, "performance": null },
      "results": [
        {
          "check": "geo-ai-bot-gptbot",
          "status": "fail",
          "message": "GPTBot is blocked in robots.txt",
          "recommendation": "Remove the Disallow rule for GPTBot to allow ChatGPT to index your content"
        }
      ]
    }
  ]
}

See the Checks Reference for the full list of checks and what each result means.

How It Works

The MCP server is a thin TypeScript wrapper around the CLI binary:

  1. The AI assistant sends a tool call with the parameters above
  2. The MCP server builds the equivalent CLI command (geodaddy <url> [flags])
  3. It spawns the binary as a subprocess and captures stdout/stderr
  4. The output (JSON or beauty text) is returned to the AI assistant

Binary resolution order:

  1. mcp/bin/geodaddy — downloaded by the postinstall script
  2. target/release/geodaddy — built locally via cargo build --release
  3. Error with installation instructions if neither is found

Execution limits: 120-second timeout per call, 10 MB output buffer. Large site crawls (--max-pages with high values) may approach the timeout.

Troubleshooting

Tool doesn't appear in Claude Desktop

Restart Claude Desktop after editing the config file. Check that the JSON is valid (no trailing commas).

"geodaddy binary not found" error

The postinstall download may have failed. Try reinstalling:

npm install -g geodaddy-mcp

Or build from source and ensure target/release/geodaddy exists:

cargo build --release

Chromium download prompt

enable_js and vitals both require a headless browser. On first use, Chromium (~150 MB) is downloaded automatically. This is a one-time download — subsequent calls reuse the cached binary.

Analysis times out

The server allows up to 120 seconds per call. For large crawls, reduce max_pages or run the CLI directly for long-running jobs.

Prebuilt binary not available for my platform

Prebuilt binaries are provided for macOS (Intel & Apple Silicon), Linux (x86_64 & arm64), and Windows (x86_64). For other platforms, build from source:

cargo build --release
PreviousChecks Reference