Start free. Scale when you need a team watching with you.
The core detection engine is open source and free forever. Pay only when you want a hosted dashboard, a team, or an AI that fixes things for you.
Open source
Always freeThe MIT-licensed SDK, running entirely in your own infrastructure.
Get started- Structural, semantic, and availability drift detection
- Learns a behavioral baseline from real traffic — no docs needed
- Local JSON baseline and incident log
- Unlimited dependencies and requests, self-hosted
- Community support on GitHub
Cloud
Self-serveConnect the SDK to a hosted dashboard — every log and detection across your codebase, in one place.
Start free trial- Everything in Open source, plus:
- Hosted dashboard across every service and environment
- Centralized baseline, synced from every repo running the SDK
- Team seats with shared incident history
- 90-day incident retention
- Email support
Enterprise
Full serviceAgentic auto-fixing and a dedicated AI assistant for teams that want drift handled, not just reported.
Contact us- Everything in Cloud, plus:
- Agentic auto-fix — proposes and applies migrations automatically
- Dedicated AI chatbot trained on your dependency history
- SSO / SAML and audit logs
- Custom data retention
- Deployment flexibility — VPC, on-prem, or BYOC
- Dedicated support and onboarding
Open core
The detection engine that actually catches drift is free and MIT-licensed, forever. You're never locked out of the thing that protects you.
Usage-based cloud
The hosted dashboard scales with how much of your codebase you connect — pay for visibility, not for seats you don't use.
AI where it earns its keep
Agentic auto-fixing and the chatbot are priced separately because they do real work — proposing and applying fixes, not just displaying data.
Frequently asked questions
Is the open source SDK really free forever?
Yes. The detection engine — the part that observes traffic, learns a baseline, and catches drift — is MIT-licensed and runs entirely on your own infrastructure. We never meter or gate that.
What does the Cloud plan actually add?
The SDK writes locally by default. Cloud points it at a hosted API instead, so every service running Undrift reports into one dashboard — baselines, incidents, and migration tips across your whole codebase, not just one repo.
How is usage measured on Cloud?
By monitored requests across all connected services, the same way the SDK already counts them locally. You can see projected usage in the dashboard before it affects your bill.
What's the difference between a migration tip and agentic auto-fix?
Every plan can detect drift and suggest a fix. Agentic auto-fix is the Enterprise capability where an AI agent applies that fix for you — opening a PR or patching the response in-flight, with you still able to review and roll it back.
Can we self-host the Cloud dashboard instead?
That's part of the Enterprise plan's deployment flexibility — VPC, on-prem, or bring-your-own-cloud, with the same dashboard and data never leaving your network.
Most tools tell you your server is down.
Undrift tells you your dependencies are lying to you.
No docs, no changelog, no vendor announcement. Caught purely from behavior — and fixed before your users notice.