Every component below was last checked under a minute ago, from the Cloudflare edge. Times are UTC. This is the DEVELOPMENT stack; it is not what customers use.
Not reporting: Agent runtime, 3D simulation. These are excluded from the headline above rather than folded into it — a surface we cannot measure is not evidence either way.
Platform
Web appOperational
The consumer app and marketing site, fetched from the edge as a visitor would. Any answer below 500 counts as reachable — the app root redirects into a locale, so a 2xx is not the healthy shape here.
Web server runtimeOperational
The web app's server-side API proxy route. It answers 401 to an anonymous probe, which proves the worker is running route handlers rather than only serving assets. It is not a test of the hop itself.
APIOperational
api.commandagi.com liveness — the worker that serves every client and agent.
API data planeOperational
The model catalog, served from D1 — proves the API can read its database, not merely that it booted.
Monitoring pipelineOperational
The API's own status snapshot and its age. A stale snapshot means the platform cron has stopped and every internal component below is frozen at its last reading.
Status pageOperational
HTTP 200
Realtime sessionsOperational
Durable Object round-trip 36ms (first contact 1160ms)
Compute
Agent runtimeNot reporting
1 agent VM(s) registered, none has streamed yet
VM provisionerOperational
heartbeat healthy
3D simulationNot reporting
8 sim instance(s) registered, none has streamed yet
CommandAGI modelOperational
/healthz ok v0.1.0, escalation live
CAD kernel (OpenCASCADE) warm poolOperational
idle — scaled to zero by design; the next call pays a ~75 s cold start
Editor services
CAD kernelOperational
aws/main-project-eval ok (402 ms); gcp/hello-commandagi-threed ok (250 ms)
Draw rendererOperational
/healthz ok (392 ms)
CAD kernelOperational
/healthz ok (214 ms)
Docs exportOperational
booted, engine still warming (cold start) (381 ms)
Sheets engineOperational
/healthz ok (140 ms)
Data queriesOperational
/healthz ok (381 ms)
Decks exportMajor outage
no such function — every call from this worker fails (lambda invoke HTTP 404 — Function not found: arn:aws:lambda:us-east-2:911797456933:function:decks-eval:$LATEST)
Music rendererOperational
/healthz ok (506 ms)
Video rendererOperational
/healthz ok (639 ms)
Circuit simulationOperational
/healthz ok (747 ms)
3D-print slicerOperational
/healthz ok (531 ms)
CNC toolpathsOperational
/healthz ok (534 ms)
Assets
Editor kernels (R2)Operational
The public bucket serving the OpenCASCADE and PlaneGCS WASM kernels. Endpoint reachability only — there is no health contract on a bucket.
Public asset library (R2)Operational
The public bucket serving library models and marketing media. Endpoint reachability only.
Hosting
Agent-built sitesOperational
Wildcard routing and TLS on *--dev.commandagi.app. A 404 from the dispatcher is the healthy answer here: it proves DNS, the certificate, the route and the tenant dispatcher are all live, which no 2xx from a specific site could.
IMAPMajor outage
imap.commandagi.com:993 — TLS handshake plus the IMAP4rev1 greeting.
SMTP submissionMajor outage
smtp.commandagi.com:587 — the STARTTLS submission port agents send mail through.
SMTP (implicit TLS)Major outage
smtp.commandagi.com:465 — the implicit-TLS port mail clients use.
Upstream providers
Vendors CommandAGI runs on. Their state is read from their own status feeds and is reported here for context — it never moves the headline above, because it is not ours to fix.
CloudflareOperational
Workers, D1, KV, R2, Durable Objects
Amazon Web ServicesOperational
cloud computers, ECR, Lambdas · us-east-2
Google CloudOperational
GCE provisioner · agent VM pool
StripeOperational
Billing & payments
GitHubOperational
Releases, CI, OAuth sign-in
Google WorkspaceOperational
Google OAuth sign-in
OpenAIOperational
LLM inference · voice (STT/TTS realtime)
AnthropicOperational
LLM inference
Fireworks AIOperational
Hosted open-weight inference · voice STT
TwilioOperational
Phone numbers · SMS · voice calls
Recent incidents
Editor kernels (R2): degraded performance16 Oct 2026, 19:03 UTC
answered HTTP 404 in 1532 ms
Twilio: degraded performance16 Oct 2026, 17:35 UTC
SMS: degraded performance
Status page: major outage16 Oct 2026, 17:01 UTC
unreachable: The operation was aborted
Public asset library (R2): degraded performance16 Oct 2026, 16:59 UTC
answered HTTP 404 in 1761 ms
API: degraded performance16 Oct 2026, 16:17 UTC
healthy in 1844 ms
Monitoring pipeline: degraded performance16 Oct 2026, 15:09 UTC
snapshot 0 s old
Public asset library (R2): degraded performance16 Oct 2026, 11:29 UTC
answered HTTP 404 in 1548 ms
Public asset library (R2): degraded performance16 Oct 2026, 10:49 UTC
answered HTTP 404 in 2257 ms
API: degraded performance16 Oct 2026, 10:46 UTC
healthy in 1785 ms
Public asset library (R2): degraded performance16 Oct 2026, 07:09 UTC
answered HTTP 404 in 1615 ms
Recent releases
devmain@ecb595616 Oct 2026, 17:01 UTC
devmain@ecb595616 Oct 2026, 16:57 UTC
devmain@ecb595616 Oct 2026, 16:56 UTC
devmain@d9b6fa116 Oct 2026, 06:02 UTC
devmain@d9b6fa116 Oct 2026, 05:59 UTC
devmain@d9b6fa116 Oct 2026, 05:57 UTC
devmain@8fd95fb16 Oct 2026, 03:09 UTC
devmain@8fd95fb16 Oct 2026, 03:05 UTC
devmain@8fd95fb16 Oct 2026, 03:04 UTC
devmain@3be5dab16 Oct 2026, 00:22 UTC
devmain@3be5dab16 Oct 2026, 00:18 UTC
devmain@3be5dab16 Oct 2026, 00:17 UTC
devmain@8975fd415 Oct 2026, 15:59 UTC
devmain@8975fd415 Oct 2026, 15:56 UTC
devmain@8975fd415 Oct 2026, 15:55 UTC
Subscribe
Get an email when a major incident opens or clears. Double opt-in; one click to leave. Machines are better served by the Atom feed, the JSON API or a status badge.
Live capacity
What is running right now, and how much more could start. Read under a minute ago.
| Machine | What runs | Region | Running | Can start now |
|---|---|---|---|---|
| Lite Computer | vm.cpu-2 | us-east-2 | 0 | 16 |
| Standard Computer | vm.cpu-4 | us-east-2 | 0 | 8 |
| Performance Computer | vm.cpu-8 | us-east-2 | 0 | 4 |
| GPU Computer | vm.gpu-l4 | us-east-2 | 0 | 1 |
| Robot Simulation | sim.simworld-1 | us-east-2 | 0 | 16 |
| The Matrix | sim.the-matrix-1 | us-east-2 | 0 | 1 |
| Transcripts | svc.faster-whisper-1 | us-east-2 | 0 | 8 |
| Human Pose | svc.vitpose-1 | us-east-2 | 0 | 2 |
| 3D Objects | svc.mask2former-1 | us-east-2 | 0 | 1 |
| Scene Graph | svc.commandagi-000 | us-east-2 | 0 | 16 |
| Semantic Video | svc.yolo-world-1 | us-east-2 | 0 | 1 |
| VLA Policy | svc.smolvla-1 | us-east-2 | 0 | 1 |
| svc.scene-splat-1 | svc.scene-splat-1 | us-east-2 | 0 | off |
| Browser | svc.chromium-1 | us-east-2 | 0 | 16 |
| Browser (instant) | svc.cf-browser-1 | us-central1 | 0 | 8 |
| svc.opencascade-1 | svc.opencascade-1 | us-central1 | 0 | 20 |
| Android embodiment | mobile.browserstack-android-1 | us-central1 | 0 | 4 |
| iOS embodiment | mobile.browserstack-ios-1 | us-central1 | 0 | 4 |
| mobile.appetize-android-1 | mobile.appetize-android-1 | us-central1 | 0 | 8 |
| mobile.appetize-ios-1 | mobile.appetize-ios-1 | us-central1 | 0 | 8 |
Every machine above draws on these shared ceilings, so they are what actually limits a launch — not a per-machine number.
| Budget | Quota | Used |
|---|---|---|
| gce-us-central1-vcpu (assumed) | CPUS_ALL_REGIONS | 0 / 32 vcpu |
| gce-us-central1-l4 | NVIDIA_L4_GPUS | 0 / 0 gpu |
| ec2-us-east-2-standard-vcpu | L-1216C47A | 0 / 32 vcpu |
| ec2-us-east-2-g-vcpu (assumed) | L-DB2E81BA | 0 / 8 vcpu |
| usd-per-hour | — | 0 / 12 $/hr |
How this page is measured
Checks marked measured from outside are HTTP or raw TCP probes run by this worker, which shares no database, no code path and no deploy with the systems it reports on. Checks marked reported by the API come from the platform’s own snapshot — stronger detail, weaker independence, and shown as such. Availability counts any answer, fast or slow, as available over the last 90 days; days we did not measure are excluded from the figure and drawn grey in the bar rather than green.