Install Guardrly MCP Server for Claude and Cursor

Guardrly installs as an MCP server that routes AI agent API calls through a monitored make_http_request tool. The installer creates your API key, saves your HMAC secret locally, and updates Claude Desktop or Cursor configuration automatically.

Quick Install (Recommended)

Mac / Linux

Option 1 — One-line installer (recommended):

curl -fsSL https://guardrly.com/install.sh | bash

Option 2 — Manual install:

brew install pipx && pipx install guardrly

Windows (PowerShell)

irm https://guardrly.com/install.ps1 | iex

The installer will:

1. Check Python 3.11+ is installed
2. Install the Guardrly package
3. Detect Claude Desktop or Cursor
4. Inject the MCP config automatically
5. Log in to your account and save API key + HMAC Secret automatically

What the Installer Configures

Step	What happens	Why it matters
Python check	Verifies Python 3.11+	Guardrly runs as a local Python MCP server
Package install	Installs guardrly with pipx	Keeps the CLI isolated from other Python projects
Account login	Creates or retrieves API credentials	Connects local monitoring to the Guardrly dashboard
HMAC secret	Saves a signing secret locally	Authenticates log uploads and prevents request tampering
MCP config	Adds Guardrly to Claude Desktop or Cursor	Routes AI agent HTTP calls through the monitored tool

Manual Installation

Step 1: Install the package

brew install pipx && pipx install guardrly

Step 2: Get your API key and HMAC Secret

If you used the one-line installer above, your credentials were saved automatically — skip to Step 3.

For manual installation only: go to app.guardrly.com/settings → MCP Configuration to copy your API Key and HMAC Secret.

Step 3: Add to Claude Desktop config

• Mac: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "guardrly": {
      "command": "python3",
      "args": ["-m", "guardrly"],
      "env": {
        "GUARDRLY_API_KEY": "gly_your_key_here",
        "GUARDRLY_API_URL": "https://api.guardrly.com",
        "HMAC_SECRET": "your_hmac_secret_here"
      }
    }
  }
}

Step 4: Restart Claude Desktop or Cursor

The make_http_request tool will appear in your Agent's tool list.

Verify Installation

Ask your Agent:

"Use make_http_request to GET https://httpbin.org/get"

Then check your Dashboard at app.guardrly.com/logs. You should see a new log entry within 60 seconds.

If the log does not appear, confirm that your AI tool was restarted, your API key is present in the MCP config, and the dashboard is filtering the current time window.

What Happens After Installation?

Once installed, Guardrly monitors AI agent API calls with PII scrubbing, alert rules, and platform-specific detection for Shopify and Meta Ads. For the strategy behind these safeguards, read the MCP server monitoring guide.

Requirements

• Python 3.11 or higher
• Claude Desktop, Cursor, or any MCP-compatible AI tool
• Guardrly account (free at app.guardrly.com)