Platform
Four surfaces, one platform
One CLI and one API: deploy from a ref, wake from anything, run in a real microVM, and see exactly what happened.
deploy · main
✓3f9c2e1SBOM · provenance
✓bld_9k2fsecret scan · 0
✓dep_4q7xrunning
✓pr-42.hello.apps.voopr.dev
Deploy
From a Git ref to a URL, one shot
Point Voopr at a repository and a ref. Every build produces an SBOM and signed provenance, secrets are scanned before the image is sealed, and egress is denied until you allow it.
- →Immutable builds keyed by commit — redeploy is a pointer move, not a rebuild.
- →Every pull request gets pr-N.<app>.apps.voopr.dev automatically.
- →Custom domains, routes and edge rules sit in front of any app.
- →Rollback is instant: previous deployments keep their snapshots.
wake sources
wake wk_2f8a340 ms · cold
0restore 214 ms340 ms
Wake
Snapshots that come back in a blink
When an app goes idle its microVM is snapshotted to local NVMe and parked at zero. Anything that can address the app can wake it again.
- →HTTP requests, cron ticks, queue messages and delayed tasks are all wake sources.
- →Restore p50 is 214 ms, p99 under 350 ms on the priority pool.
- →Queues ship with dead-letter handling and visibility timeouts.
- →Reserved warm capacity skips the restore step entirely.
runtimes
✓node24handler.js
✓python312handler.py
✓go123main.go
DATABASE_URLsealed
Run
Hardware isolation for every handler
Each app gets its own Firecracker microVM with its own kernel on bare metal — sync or async invocation, stateless by design.
- →node24, python312 and go123 runtimes, or bring a container image.
- →Up to 16 vCPU and 32 GB per microVM on Scale.
- →State lives in your Postgres, object store or KV — URLs sealed at rest.
- →Secrets are injected at wake and rotatable from the CLI or console.
observability
✓hello · GET /200 · 340 ms
✓image-resize · POST /resize200 · 212 ms
✓nightly-etl · cron200 · 4.1 s
alerts0 firing
Observe
A trace for every invocation, a timeline for every wake
Logs stream as they are written, metrics are per app and per version, and every wake keeps the timeline that produced it.
- →Streamed logs with structured fields and live tailing from the CLI.
- →Wake timelines split restore, boot-resume and handler time.
- →Alerts on error rate, wake latency and queue depth.
- →Audit log of every write, exportable on Scale.