HostTracker MCP ServerVerifiedFeatured

What is the HostTracker MCP server

HostTracker has run external website monitoring since 2004 for more than 500,000 sites. This is its official MCP server: a hosted, streamable-HTTP endpoint that lets an AI assistant operate your monitoring account in conversation. Nothing to install.

What the assistant can do

• Check anything right now from 300+ global checkpoints: HTTP/S, ping, TCP port, DNS, traceroute, WHOIS, blacklist (DNSBL), Google Web Risk, page speed and crawl, with per-location results.
• Manage monitors: create, edit, pause and resume; pin locations; see what is down and why.
• Read uptime and incidents: uptime summaries, results history, incident timelines.
• Schedule maintenance windows so planned work does not page anyone.
• Manage contacts and subscriptions: who gets alerted, on which channel, for which monitors.
• Wire webhooks and redeliver failed deliveries.
• Publish status-page incidents and updates.
• Generate reports and drive long-running jobs.

65 tools cover the full HostTracker API v2, plus a guarded generic door (describe_api / api_request) for anything without a dedicated tool.

Connect

1. Mint an API token at https://www.host-tracker.com/integrations/api and choose the scopes the assistant should have (check, monitor, contact, subs, webhook, statuspage, account).
2. Add the server to your client with Authorization: Bearer <token> (config block above). Claude Code, Cursor, VS Code, Windsurf and any stdio-only client via mcp-remote are covered in the client guide.
3. Ask: "Run an http check on example.com from Europe and North America", "Which of my monitors are down?", "Pause staging until tomorrow morning", "What was my uptime last month and what caused the longest outage?"

Safety

The server is stateless and forwards every call under your own token, where quota, ownership and rate limits are enforced. Writes to the account itself (profile, email, billing) are refused regardless of scopes; payments, plan changes, passwords and token minting are not on the API at all. Destructive bulk operations validate first and require an explicit confirming call. Anything a checked target controls (error strings, WHOIS records, DNS TXT, response excerpts) is fenced as untrusted content before it reaches the model.

Links

• Landing page: https://www.host-tracker.com/mcp-server
• Repository (client guide, security policy, registry manifest): https://github.com/HostTracker/mcp
• Official MCP Registry entry: io.github.HostTracker/hosttracker