Feature
Edge-case mode
The inputs your fixtures never think to include — still perfectly valid, just drawn from the corners where naive code breaks.
Most test data sits comfortably in the middle of every format: a five-letter name, a date in the middle of the month, an identifier with nothing unusual about it. That is exactly the data that never finds your bugs.
Edge-case mode flips that. Set edge=true on any generator and every record is drawn from the rare-but-valid corners of its format — the values that are entirely legal, yet break code that quietly assumed the common case.
Valid, never invalid
Edge-case mode never produces malformed data. Every record still passes the format’s own validation — correct checksums, legal lengths, real structure. It only steers which valid value you get, biasing toward the corners instead of the comfortable middle. (If you want deliberately-wrong data — bad check digits, broken formats — that is what Invalid mode is for.)
What counts as a corner
It depends on the generator. For an email, that is the longest RFC-valid local part, plus-addressing tags, and the rarer legal characters. For a person name, the longest and shortest names, diacritics, and apostrophes. For a national identifier, the dates and serial ranges most likely to trip up a hand-rolled parser. Each call draws one corner at random, so a batch spreads across them.
Why it catches bugs
Column too short for the longest legal name. A date parser that assumes two digits. A regex that forgot apostrophes are allowed. A checksum routine that is right on average but wrong at the boundary. These never show up against average data — and they are exactly what ships to production untested.
Pins still win
Edge-case mode only steers the dimensions you leave open. Pin a domain, a pattern, or a country and that choice is honoured — edge bias applies to everything else. You stay in control of the shape; it just pushes the rest toward the corners.
Try it
Add edge=true to any generator request:
curl "https://api.real-fake-data.com/v1/person-name?edge=true"Edge-case mode is included on the Pro and Self-hosted plans.
From the blog
- Testing the unhappy path with edge and invalid modes → How edge and invalid modes exercise the boundary values and rejection paths your happy-path tests never reach.
- Introducing extreme mode: correct data in a hostile encoding → The next step beyond edge — correct values wrapped in whitespace, invisible characters, and homoglyphs.
Put it to work
Edge-case mode is available on Pro and Self-hosted. Try it live in the Playground, or pick a plan.