Pricing
The browser-local validator stays free. Pro and Enterprise are planned for teams that want migration help, CI and a shared registry.
The full browser-local validator and schema comparison, for individuals and open source.
No account required.
Migration help and automation for teams that ship API changes every week.
Not billable yet. Nothing is charged.
For organisations with security review, residency and support requirements.
Scoped with you before anything is agreed.
| Feature | Free Community | Pro / Team | Enterprise |
|---|---|---|---|
| Core checks | |||
| Client-side validator and comparison | Included | Included | Included |
| JSON Schema Draft-07 and 2020-12 | Included | Included | Included |
| Documents stay in your browser tab | Included | Included | Included |
| Local CLI | Planned | Planned | Planned |
| AI and automation | |||
| Claude AI Migration Explainer | Not included | Beta | Beta |
| Breaking-change release notes | Not included | Planned | Planned |
| GitHub Action CI integration | Not included | Planned | Planned |
| Internal schema registry | Not included | Planned | Planned |
| Security and operations | |||
| SSO / SAML | Not included | Not included | Planned |
| Self-hosted on-premise / VPC | Not included | Not included | Planned |
| Contractual SLA | Not included | Not included | Planned |
| Custom schema dialects | Not included | Not included | Planned |
Planned means on the roadmap and not yet released; Beta means you can try it now. Pro and Enterprise cannot be purchased yet: prices are the planned list price, and joining a waitlist creates no account and no charge.
No. Validation and comparison run in your browser, in a Web Worker, using the Ajv engine. Your schemas and payloads are never sent anywhere.
The optional Claude explainer is the only feature that makes a network request, and only when you click it and supply your own Anthropic key (or choose the offline mock). It sends the breaking findings the engine already extracted: JSON Pointer paths, rule names and messages. Those can include property names, so review them if names are sensitive.
Schema Labs never receives your findings or your key: the request goes from your browser straight to the Anthropic API. Anthropic’s commercial terms for its API say that customer inputs and outputs are not used to train its models by default. Check Anthropic’s current terms for your account and plan, because they govern that request.
The rules decide what changed and how severe it is. They are deterministic: the same two schemas always produce the same findings, and Claude cannot add, remove or re-rate one. They are also conservative, so some changes are flagged for review rather than judged.
Claude acts as an editor after the fact. It explains why a finding breaks consumers, drafts an adapter and writes a regression test. Treat that output as a draft to review.