Schemasift: Data Drift

Schemasift compares two JSON Schemas and tells you which field changes will break the systems already reading your data before you ship the new version. Work on your own contract in five steps: 1. Paste or open your current schema as the baseline, then your proposed revision as the candidate. Draft 2020-12 parsing enforces explicit size, depth, and node limits and reports malformed syntax instead of crashing. 2. Read the directional verdict: backward (new readers accept old documents), forward (old readers accept new documents), and full. Every finding lands at a precise JSON Pointer with a before/after value and a severity up to blocker. 3. Attach real JSON documents as samples and replay them against both contracts. Schemasift reports how many pass each side and names the exact keyword and pointer for each failure. 4. Set contract context, owner, consumers, documents per day, and the policy you enforce, so the risk score reflects your real blast radius. 5. Export a deterministic JSON or Markdown report. Identical inputs produce identical bytes and the same SHA-256, so it attaches to a pull request as evidence. Prefer a running start? Three complete projects (checkout events, device telemetry, inventory inference) load in one tap and are yours to edit or clear. What it is not: Schemasift never fetches external reference targets, does not claim complete RFC 8785 canonicalization, and treats a passing replay as evidence, not proof. Regex, composition, and dynamic references are flagged for manual review. Privacy: every comparison runs locally and is initiated by you; schemas and samples stay in the app's local container, leaving only through the system share sheet