Report · 19 November 2026 · published monthly
The state of MCP servers, November 2026
We connected to every company MCP server in the official MCP Registry the way an AI app does, twice, at least ten minutes apart. Most answered. Some never did.
1 in 20company MCP servers in the official registry failed both of our checks.
What broke
For each broken server, the answer it gave the second time. Most of these are not subtle bugs. The address itself is gone, or nothing is listening behind it.
| Cause | Servers |
|---|---|
| The address no longer exists (DNS) | 27126% of the broken servers |
| The server crashed or is unreachable (5xx) | 25524% of the broken servers |
| Nothing at the listed path (404) | 24624% of the broken servers |
| No answer within 10 seconds | 10710% of the broken servers |
| A website answers, not MCP (405, HTML) | 808% of the broken servers |
| Certificate or TLS error | 747% of the broken servers |
| Refuses a standard client (other 4xx) | 71% of the broken servers |
| Other | 1<1% of the broken servers |
What each one means, and the fix
- The address no longer exists (DNS). The hostname in the registry entry doesn't resolve, at Cloudflare's and Google's public resolvers either. Usually a domain that lapsed, a subdomain someone deleted, or a preview hostname that was never meant to last.
Point the name back at your server, or publish a new registry version with the address that works today. If you retired the server,
mcp-publisher status --status deletedtakes it out of the registry's listings. - The server crashed or is unreachable (5xx). The address answers, but the app behind it doesn't: a 502 or 503 from the proxy in front, or Cloudflare's 530 when the tunnel or origin behind it is gone (61 of these). The host's error page names the layer that failed. Restart or redeploy, then add the MCP route itself to your uptime check, not just the home page.
- Nothing at the listed path (404). The domain works, the MCP route doesn't. It moved, for example from /mcp to /api/mcp, or /sse went away in a switch to Streamable HTTP, or a redeploy left it out. Make the URL in your registry entry match the route you serve, then publish a new version.
- No answer within 10 seconds. Timeouts and refused connections, both times. Often a server that sleeps and wakes too slowly, sometimes a firewall. Answer initialize before loading anything heavy, or keep one instance warm. Our check waits 10 seconds.
- A website answers, not MCP (405, HTML). The address serves a web page or refuses the POST every MCP request is. Usually the site's catch-all route answering because the MCP route isn't deployed, or a registry entry that lists the home page. Serve MCP on its own path that accepts POST, and list that exact URL.
- Certificate or TLS error. The certificate expired, was issued for another name, or is self-signed, so a client refuses to connect at all. Renew it, or put the server behind a host that renews certificates for you.
The rest (8) failed in other ways, such as asking for a session ID on the very first request.
Servers that only speak the newest protocol
24 servers turned down our handshake because they only speak the newest version of the protocol, 2026-07-28. They're running, so we count them as working. But not every client can reach them yet. The current release of the official TypeScript SDK (1.32.2) doesn't speak that version, and the Python SDK (2.3.0), which speaks both, listed the tools of 16 of them in our test. The others asked for a header it doesn't send, or answered with a tool list it rejects. If you run a server like this, accepting the older handshake too keeps today's apps working.
Test yours
From a terminal, send the first message every client sends. Put your server's address in place of the example:
curl -i https://your-server.example/mcp \
-H 'Content-Type: application/json' \
-H 'Accept: application/json, text/event-stream' \
-d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"check","version":"1.0"}}}'
A working server answers 200 with JSON that includes "protocolVersion", either as the body or on the data: line of an event stream. Anything else is what an app sees too. To walk through the whole exchange, run the official MCP Inspector: npx @modelcontextprotocol/inspector.
Or test it in your browser with our free MCP server test, the same check this report used. For a human look, send us the address: we tell you what fails and why, free, within 24 hours. If you have an error message instead, our MCP connector looks it up in our fix guides from inside Claude or any MCP client.
How we measured
The list. Every server in the official MCP Registry on 19 November 2026 that lists a remote address: 45,531 entries. We kept the ones published under a company's own domain, not a GitHub account or a free hosting subdomain, that answer without the user's own API key. Then one address per host, leaving out platforms that host thousands of servers for others, such as Smithery, Apify and Glama. That leaves 21,180 company servers.
The check. The first two steps any client takes: initialize, then list the tools, over Streamable HTTP, with 10 seconds for each. For servers on the older SSE transport, opening the event stream. We never call a tool.
Twice. A server counts as broken only if both checks failed, at least 10 minutes apart. When the address didn't resolve, we also asked Cloudflare's and Google's public DNS, and counted it only if both failed too.
What counts as working. Any real answer from the server, including one that asks for a key or a payment (401, 402, 403), limits our rate (429), blocks our region (451) or speaks only the newest protocol version.
Left out of the count. 102 addresses on throwaway developer tunnels (trycloudflare.com, ngrok and similar) and 99 servers switched off on purpose: 410 Gone, or a host saying the deployment is disabled or paused. They failed too, but nobody expects them to answer.
Limits. We check from one place, in Canada, so a server that blocks our network, or was down for exactly those minutes, counts as broken. And a server that passes can still fail when a tool is called, which we don't test.
No names. We don't publish which servers failed. The counts behind this page are in 2026-11.json. We wrote the checking scripts and this page with AI tools, including Claude. Every number comes from the check logs.
Next edition: December 2026.
Most of these breaks take minutes to fix, once someone knows they're there.