// the find
mathiasbynens/small
Smallest possible syntactically valid files of different types
A grab-bag of the smallest possible byte-valid files across dozens of formats — HTML, PNG, ZIP, PE, WebAssembly, even Mach-O and DOS executables. It's for people building or testing parsers, decoders, or file-type sniffers who need minimal fixtures rather than production-sized samples.
Genuinely exhaustive breadth — over 100 formats from mainstream (JSON, SVG, MP4) to obscure (INTERCAL, Malbolge, TADS-3). Each file is a real, working minimal example rather than a stub, which makes them directly usable as parser test fixtures or magic-byte references. Public domain (CC0), so there's zero licensing friction to vendor these into a test suite.
No documentation of *why* each file is minimal or what was stripped to get there — you can't learn the format spec from this, only copy the bytes. No automated tests or CI validating that these files still parse correctly on their respective decoders (last CI config dates back years), so silent bit-rot is possible. It's a flat, unversioned collection — no way to pull just the subset you need without cloning the whole repo, and PR-driven curation means quality/consistency varies by contributor.