MCP Server for AI Agents

Connect Claude, Cursor, Codex, VS Code and any other Model Context Protocol client to your Amplify account. Your agent can then answer questions like "which of my servers have open NGINX CVEs?" or "why did 5xx errors spike on web1 last night?" straight from your monitoring data.

Server URL & authentication

• URL: https://amplify.getpagespeed.com/mcp (Streamable HTTP, hosted, nothing to install)
• Auth: your account API key as Authorization: Bearer YOUR_API_KEY (or an X-Api-Key header). Find it on the API Keys settings page.
• Read-only: every tool only reads data. Nothing in your account can be changed through MCP.

Setup

Claude Code

claude mcp add --transport http --scope user amplify \
    https://amplify.getpagespeed.com/mcp \
    --header "Authorization: Bearer YOUR_API_KEY"

Cursor

Add to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "amplify": {
      "url": "https://amplify.getpagespeed.com/mcp",
      "headers": { "Authorization": "Bearer YOUR_API_KEY" }
    }
  }
}

VS Code (GitHub Copilot)

Add to .vscode/mcp.json or your user MCP configuration:

{
  "servers": {
    "amplify": {
      "type": "http",
      "url": "https://amplify.getpagespeed.com/mcp",
      "headers": { "Authorization": "Bearer YOUR_API_KEY" }
    }
  }
}

OpenAI Codex CLI

Add to ~/.codex/config.toml and export AMPLIFY_API_KEY in your shell:

[mcp_servers.amplify]
url = "https://amplify.getpagespeed.com/mcp"
bearer_token_env_var = "AMPLIFY_API_KEY"

Other clients

Any client that supports remote Streamable HTTP servers with custom headers works with the URL and header above. Web connectors that only support OAuth sign-in (claude.ai, ChatGPT) are coming next.

Tools

Every host argument accepts a host UUID or its hostname.

Tool	What it answers
get_security_overview	Every host ranked by NGINX CVEs, gixy config findings and PHP CVE exposure, in one call
get_host_security	One host's CVEs, config findings with suggested fixes, and PHP-FPM verdicts with upgrade commands
list_hosts / get_host	Inventory: online state, agent version, server kinds, tags
list_host_components	Detected NGINX, PHP-FPM, MySQL and Varnish instances with versions
list_host_metrics	Metric names a host reports
get_metric_series	Up to 10 metrics over a window; a compact min/max/avg/last summary by default, raw points on request
get_latest_metric	Current value of one metric
list_alert_rules / list_recent_alerts	Configured alerts and what fired recently (with hostnames, filterable by time window)
list_websites / get_website / list_website_checks	Uptime, response time, HTTP/3, TLS and certificate checks
get_account	Plan, host limit, retention and rate limit

Things to ask

• "Which of my servers need security attention, and what should I fix first?"
• "Show me the gixy findings on web1 and write the NGINX config changes that fix them."
• "Compare request rate and 5xx errors on all hosts over the last 24 hours."
• "Did any website go down this week? How long, and what was the error?"
• "Which hosts run an outdated Amplify agent or an end-of-life PHP version?"

Limits & data

MCP tools are built on the REST API and share its rules: 120 calls per minute per API key (REST and MCP draw from the same budget), time windows clamped to your plan's retention, and the same data your dashboard shows. Your agent sees your monitoring data; review your AI client's data-handling terms before connecting production accounts.