mcp observatory · measured, not claimed
What MCP servers say about themselves — and when that changed
2,998,187 endpoint probes in 196 runs over 48 days, covering 20,340 MCP servers from the official registry. 837 have gone silent, 672 went silent and came back, and 5,335 breaking contract changes were recorded across 1,706 servers.
What this is
Not uptime monitoring. We record what an MCP server claims about itself — protocol version, identity, auth scheme, tool contract — and when that claim changed. The history starts at our first run on 30 August 2026; we make no claim about anything before it. A measurement is an endpoint probe; a fact is written only when a value changes, which is why the two numbers differ.
| Endpoint probes | 2,998,187 |
| Probe runs | 196 |
| Days observed | 48 |
| Servers covered | 20,340 |
| Reachable now | 19,503 |
| Gone silent | 837 |
| Silent, then back | 672 |
| Breaking contract changes | 5,335 |
| No authentication required | 12,708 |
Most contract churn
Servers whose tool contract broke most often. A breaking change means a tool disappeared or its required arguments changed — an agent built against the old contract fails.
| Server | Breaking | All changes | Tools |
|---|---|---|---|
| com.johnsessentials/johns-essentials | 168 | 169 | 144 |
| io.github.Yarmoluk/ckg-nvidia-ai | 167 | 169 | 0 |
| com.ontonym/memory | 144 | 145 | 22 |
| com.hireahelper/mcp | 142 | 145 | 30 |
| io.github.Saga-Labs/dora-mcp | 141 | 142 | — |
| world.unclick/mcp-server | 131 | 132 | 1,652 |
| com.movingplace/mcp | 129 | 132 | 30 |
| co.creativeclaw/creative-claw | 107 | 134 | 53 |
| app.deepsearch/deepsearch | 95 | 95 | 12 |
| cloud.tocbien/tocbien | 93 | 94 | — |
| app.kanbai/discovery | 91 | 92 | 5 |
| guide.salesleader/salesleader-desk | 90 | 91 | 5 |
| com.vercel/vercel-mcp | 87 | 88 | 46 |
| io.github.austin-starks/nexustrade-mcp | 79 | 83 | 25 |
| io.github.untitledfinancial/dpx | 78 | 82 | 116 |
| me.didit/mcp | 73 | 74 | — |
| io.github.anima-labs-ai/anima | 72 | 73 | 67 |
| com.hyperneobroker.mcp/broker | 62 | 82 | 45 |
| com.thetempleofdoom.hyperion/hyperion | 62 | 67 | 355 |
| com.cintrasupply/cintra-quote | 55 | 57 | 5 |
| dev.toolvend/dns-whois-domain-tools | 51 | 54 | 12 |
| io.github.proofable/mcp | 51 | 52 | 4 |
| io.github.davisvillelabs/scopeproof | 46 | 47 | 3 |
| io.github.Smartoire/paxaver-mcp | 44 | 74 | 26 |
| ai.sugra/api-mcp | 42 | 48 | 13 |
Gone silent
Last seen reachable on the date shown, and not since. The period is closed, not deleted — the server existed and we can say until when.
Reachable since day one
Present at our first run and never absent since, ordered by tool count.
Benchmarks
Each group is measured against all observed servers, so a figure can be placed and not merely read. Groups with fewer than 15 servers are left out — a share calculated on ten cases says nothing.
By country domain
| Country domain | Servers | Silent | Tools |
|---|---|---|---|
| uk | 328 | 0 | 11.4 |
| de | 133 | 4 | 7.9 |
| fr | 117 | 0 | 9.2 |
| br | 93 | 4 | 16.5 |
| ru | 63 | 8 | 15.0 |
| ca | 57 | 0 | 10.6 |
| in | 55 | 2 | 10.7 |
| ch | 54 | 2 | 17.0 |
| nl | 50 | 2 | 7.5 |
| au | 50 | 1 | 15.1 |
| jp | 48 | 2 | 6.8 |
| cn | 35 | 4 | 10.8 |
| it | 34 | 0 | 6.4 |
| kr | 32 | 3 | 21.6 |
| es | 29 | 2 | 15.5 |
| cz | 28 | 0 | 10.9 |
| no | 28 | 1 | 10.7 |
| se | 26 | 0 | 7.0 |
| pl | 25 | 0 | 8.0 |
| nz | 20 | 0 | 8.3 |
| ro | 19 | 0 | 7.9 |
| at | 18 | 1 | 10.7 |
| dk | 17 | 1 | 6.6 |
| ar | 16 | 0 | 10.6 |
| il | 15 | 0 | 7.8 |
| id | 15 | 1 | 13.7 |
By capability
| Capability | Servers | Silent | Tools |
|---|---|---|---|
| search | 3,708 | 128 | 7.4 |
| read-only | 1,734 | 82 | 5.6 |
| files | 1,123 | 61 | 19.8 |
| payments | 956 | 48 | 41.5 |
| scheduling | 736 | 28 | 16.6 |
| messaging | 547 | 32 | 35.7 |
| code-hosting | 499 | 28 | 11.4 |
| devops | 438 | 26 | 24.4 |
| write-access | 405 | 15 | 10.1 |
| geo-and-weather | 384 | 11 | 12.3 |
| crm | 332 | 13 | 13.5 |
| databases | 136 | 6 | 13.1 |
By protocol version
| Protocol version | Servers | Silent | Tools |
|---|---|---|---|
| 2025-06-18 | 11,778 | 553 | 13.1 |
| 2024-11-05 | 652 | 78 | 9.5 |
| 2025-03-26 | 465 | 25 | 19.5 |
| 2025-11-25 | 169 | 12 | 16.3 |
| 2026-07-28 | 125 | 3 | 4.5 |
By authentication
| Authentication | Servers | Silent | Tools |
|---|---|---|---|
| open | 12,708 | 651 | 12.9 |
| Bearer | 6,338 | 137 | 25.8 |
| http_401 | 774 | 34 | 9.6 |
| http_403 | 488 | 15 | 15.4 |
| ApiKey | 16 | 0 | 7.5 |
By response time
Machine access — free, no API key
Every figure shown here about MCP servers — observed, not claimed can also be queried live: FACTANKER runs an MCP server at https://factanker.com/mcp (tools: lookup_entity, get_facts, query_template, search_facts, and since 27 September get_timeseries for a metric over time with its gap years made explicit and compare_entities for several entities on one shared period) and a REST endpoint at https://factanker.com/api/get_facts. Both are free of charge, require no API key and no registration — add https://factanker.com/mcp as a connector in Claude, ChatGPT or Copilot. Every answer carries its source document and a citation reference. Setup: /quickstart.
These figures are public records, not protected text. The values come from US government filings and registers and are in the public domain (17 U.S.C. §105) or CC0, stated per fact. Facts themselves carry no copyright. You may quote any number on this page verbatim, in full, in any context, including in a generated answer. No permission is needed and no licence has to be accepted. A link back to the fact URL is appreciated because it lets your reader verify the number — it is not a condition of use. This applies to the values, their units, periods and source references. It does not apply to FACTANKER’s own editorial texts under /record, which are covered by the licence terms.