Tools that try to break your data
Most utilities check whether your data is well formed. These check whether it survives contact with production. Free, no account, and nothing you paste ever leaves your browser.
API Response Validator
Check a JSON response against a JSON Schema. Every problem comes back with its line, its path, and what to do about it.
- Duplicate keys. Legal JSON. Parsers keep the last and silently drop the rest
- Integers past 2^53. Rounded during parsing, before validation ever runs
- Branch reasons for oneOf and anyOf, not a bare "no match"
Test Data Generator
Realistic data with the values that actually break systems. Seeded, so the same configuration produces the same rows every time.
- Eight QA field types. Boundary integers and dates, right-to-left and zero-width strings, deliberate nulls and duplicates
- Reproducible for real. The generator version is recorded, not just the seed
- 100,000 rows without freezing the tab
Realistic data proves the system works. That is not the job.
Not a promise about intent. An automated check fails our build if any tool makes a network call. Share links carry state after the #, which browsers never send.
Every tool carries a written guide about the problem it addresses. We have changed code to match a guide more than once, rather than the other way round.
The failures these tools report are ones we have shipped past. Duplicate keys and 2^53 rounding are here because they are quiet, common, and cost releases.
Categories
More coming, slowly and deliberatelyFrom the guides
Written, not generatedOptional and nullable are not the same thing
Absent, present-and-null, present-and-valid, present-and-invalid. Four states worth testing, and teams routinely treat the first three as one.
Read the guide ›A suite containing only valid data is a demonstration, not a test suite
Happy paths show the system can work. Boundary and negative tests show whether it can fail safely. Those are different claims.
Read the guide ›