nullstack gives every pull request its own environment — the same containers, the same database engine, the same secrets shape as production, seeded with scrubbed data.
It exists because staging is a lie. One shared environment, permanently broken by whoever merged last, is worse than no environment at all.
It is one binary and a YAML file. There is no control plane for you to run, no operator to upgrade, no agent on your cluster.
| plan | environments | price | support |
|---|---|---|---|
| solo | 3 concurrent | $0 | community |
| team | 25 concurrent | $400 / mo | shared slack, 1 day |
| scale | unlimited | $1,600 / mo | named engineer, 2 hours |
| self-host | unlimited | $24k / yr | we come to you |
Compute runs in your own cloud account and is billed by your provider, not by us. We charge for the scheduler, not for the servers.
→ reading nullstack.yaml → 6 services, 2 datastores, 1 queue → pulling scrubbed snapshot (412 MB) ......... 1.8s → building api [cached 9/11 layers] ... 4.2s → building web [cached 7/8 layers] .... 2.9s → building worker [cached 11/11 layers] .. 0.4s → provisioning postgres-16 ................... 6.1s → provisioning redis-7 ....................... 1.2s → wiring service mesh ........................ 2.0s → running migrations (14) .................... 3.7s → health checks .............................. ok environment ready in 39.4s https://pr-4192.acme.nullstack.dev teardown on merge · cost so far $0.07
We deleted our staging environment eleven days after installing this, and nobody has asked for it back.
Staff engineer, logistics platform · 310 engineers
The per-branch cost attribution ended an argument our platform team had been having with finance for two years.
VP Engineering, fintech · 90 engineers
It runs in our account, so our security review was about IAM policy rather than about trusting a vendor with data.
Head of Security, health tech · SOC 2, HIPAA
does it work with our monorepo?
?does it work with our monorepo?
Yes. Services are declared by path, and the build graph is derived from what changed in the diff — an unrelated service is not rebuilt.
what about our data?
?what about our data?
Snapshots are taken and scrubbed inside your account. Nothing leaves it. We never see your data, which is also why we cannot debug it for you.
kubernetes?
?kubernetes?
Optional. nullstack schedules onto EKS, GKE, AKS, Fargate or plain EC2. Most teams under 200 engineers pick Fargate and never think about it again.
how is this different from a preview URL?
?how is this different from a preview URL?
A preview deploys the frontend. This brings up every service, a real database with real-shaped data, the queue and the cron jobs — the parts that actually break.
what happens if you go out of business?
?what happens if you go out of business?
The scheduler binary is escrowed and the format is documented. Everything already runs in your cloud account, so the worst case is that it stops improving.
We answer within one working day. There is no SDR, no sequence, and nobody will add you to a nurture track.