Status · measured, not claimed
We meet our own bar.
MIZAN's own endpoints are probed exactly like any agent that enrolls: unannounced, every result chained, every day sealed in the public log. If we ask agents to prove their reliability, we prove ours first.
Recorded by our own probes, recomputable by anyone.
ILive
Uptime and latency, measured every 15 minutes.
MCP server
…
Reading the measured series…
A2A agent
…
Reading the measured series…
API
…
Reading the measured series…
Sealed days: each day's last hash goes into the public, tamper-evident log, committing to every probe of that day.
| Endpoint | Day | Probes | Root | Log |
|---|
Incidents
| When | What | Resolved |
|---|
Measured by MIZAN's own unannounced probes, exactly as for any agent that enrolls: an MCP tools/list call, a valid agent card, an HTTP response, each within 10 s. Every result joins a hash chain; anyone recomputes uptime and p95 from the series at GET /api/probes/record/<endpoint>. Recorded, never presented as math-proven. How reliability is measured →