Real MCP health checks
We speak MCP. Not TCP ping, not HEAD request — a full initialize handshake every 60 seconds, compared against the protocol spec.
AliveMCP
AliveMCP
91% of public MCP endpoints are dead, broken, or returning malformed JSON-RPC right now. We ping every server in every registry every 60 seconds — so you know before your users do.
Live-pinging across MCP.so, Glama, PulseMCP, Smithery, the Official Registry, and GitHub.
The problem
An April 2026 audit of 2,181 remote MCP endpoints found only 9% were healthy. The rest were returning 404s, malformed JSON-RPC, broken auth, or hanging on connect. Nobody told the authors. Nobody tells you either — until a user files a GitHub issue a month late, or worse, a registry-scan blog post lists your server as "dead" and buries it in ranking.
How it works
Every hour we crawl MCP.so, Glama, PulseMCP, Smithery, the Official Registry, and GitHub topic feeds — your public server shows up automatically within 60 minutes of being listed.
A real MCP initialize request over HTTP or SSE. We measure latency, hash the tool list, and check response shape against the protocol spec. A server that opens a socket but returns malformed JSON-RPC is flagged down.
Claim your listing for $9/mo and plug a webhook, Slack channel, or email. We page you the moment your server crosses 15 minutes down or your tool schema drifts — not a week later.
What you get
We speak MCP. Not TCP ping, not HEAD request — a full initialize handshake every 60 seconds, compared against the protocol spec.
Public page at alivemcp.com/status/your-server. Shareable link with response-time graph, incident log, and a badge you can embed in your README.
We hash your tool list every ping. The second a tool changes name, params, or disappears, we page you — before downstream agents break silently.
Plug your incident channel. We page you on downtime, latency regressions, and schema changes — not false alarms at 3am for a single blip.
Pricing
Just browsing the public dashboard? That's free — no account, no email. Paid tiers start when you want alerts, private endpoints, or compliance.
$9/mo
Indie authors with 1–3 public MCPs.
Most popular
$49/mo
Teams running internal or private MCPs.
Custom
30+ internal MCPs, SAML, compliance.
Questions
We send a real MCP initialize request over HTTP or SSE every 60 seconds and check the response shape against the protocol spec. It's not a TCP ping — we actually speak MCP. A server that opens a socket but returns malformed JSON-RPC is flagged down.
Your server used to expose tools A, B, C. A deploy accidentally removes C, or changes its input schema, and downstream agents break. AliveMCP hashes your tool list every ping and alerts you the moment the hash changes — before your users notice.
No. We crawl every public MCP registry — MCP.so, Glama, PulseMCP, Smithery, the Official Registry, and GitHub — and every discovered endpoint gets a free status page automatically at alivemcp.com/status/your-server. Claim it for $9/mo if you want custom alerts and a verified-author badge.
Author is for the MCP you already ship publicly — claim your listing, add Slack and webhook alerts, get the verified-author badge. Team is for MCPs you don't publish: private endpoints, your own status subdomain, team-level Slack, and a public status page subdomain for internal SLAs.
Enterprise customers get an on-prem container that monitors private endpoints inside your VPC and reports back over a single outbound webhook. SSO, audit log, and monthly SLA PDFs included. Email hello@alivemcp.com.
Join the waitlist. We'll email when your server's status page goes live — probably before your next deploy.
Get early access