All systems operational
Last measured (22s ago).
Website and CVE Pulse
-
Operational
echelongraph.io
The public website.
-
Operational
CVE Pulse (echelongraph.io/pulse)
The public CVE feed pages.
-
Operational
CVE Pulse data freshness
When CVE Pulse last took in an NVD change, and when CISA's KEV catalogue last answered our poll.
API and MCP
-
Operational
API (app.echelongraph.io)
The public REST API.
-
Operational
Web app (app.echelongraph.io)
The signed-in product.
-
Operational
Hosted MCP (mcp.echelongraph.io)
The keyless remote MCP endpoint, https://mcp.echelongraph.io/mcp.
Radars (last successful run)
-
Operational
EchelonGraph Radar (daily reachability cycle)
-
Operational
KEV-Exposure Radar
-
Operational
MCP Exposure Radar
-
Operational
AI-Exposure Radar
-
Operational
Shadow AI Radar
-
Operational
Exposed Data-Stores Radar
-
Operational
Exposed AI Keys & Config Radar
-
Operational
Leaked Credentials Radar
-
Operational
Email Spoofability Radar
-
Operational
Subdomain Takeover Radar
Response times, last 90 days
1m 59s
Mean time to detect
1 incident
—
Mean time to acknowledge
0 incidents
3m 59s
Mean time to resolve
1 incident
Computed from every incident on this page detected since : 1 incident (1 detected automatically by our probes, 0 reported by the team). Detect = impact start to detection; acknowledge = detection to a person taking it; resolve = impact start to impact end. 1 not yet acknowledged is left out of the acknowledge mean.
Past incidents
-
Shadow AI Radar: Partial outage
Minor impact · Shadow AI Radar · · lasted 3m 59s · detected automatically
Timeline (2 updates)
-
Resolved
Automated: Shadow AI Radar has passed every check since 2026-10-06 13:09 UTC (3 consecutive rounds). Resolved automatically; a follow-up will be posted if we learn more.
-
Investigating
Automated detection: Shadow AI Radar has failed 3 consecutive checks since 2026-10-06 13:05 UTC. Partial outage. last completed cycle 32m 0s ago (expected within 30m); reports not running.
-
Resolved
Subscribe to updates
Get an email whenever we post an incident update. We send a confirmation link first, store only your address, and every email has an unsubscribe link.
How this page is measured
Every minute a prober outside our main region requests each public service, and Google Cloud uptime checks probe them from several continents. Radars and data feeds are graded on the time of their last successful run, against the interval they are expected to run in. A component without a measurement in the last 5 minutes shows No data, never Operational. Uptime counts measured checks only, and a regional result older than 5 minutes is not counted. Latencies in our JSON are measured from the prober and include its own overhead; they are not a visitor's latency. Incidents open automatically after repeated failures and resolve when checks pass again; the team adds context to them.