Binary files are hard to read. Open one in a hex editor and you get a wall of numbers. Edit a value by hand and you can break a size field or a checksum somewhere else in the file without noticing. And the format itself usually gets described three times, once in a reader, once in a writer, once in a hex editor template, and the three copies drift apart.
EByte describes a format once. You write one schema in JSON: field order, field widths, byte order, how many records follow. EByte uses that one description to read a file into JSON you can edit, write the JSON back into a binary file, and trace a file field by field, reporting the field, the byte offset and what it expected wherever a file doesn’t match.
The Pitch in Four Numbers
- 16 MiB, the largest file decoded to JSON and written back with every byte the same.
- 1 schema for reading, writing and tracing, so they can’t drift apart.
- 0 modules outside the Go standard library, and no generated code.
- Apache 2.0, free, with the tests and recorded results on GitHub.
How It Works
The core rule is round-tripping: decode a file, re-encode the result without changes, and you get the same bytes back. Padding, reserved bytes, NULs and trailing spaces all survive, because the schema has to declare them. A mistake in the schema shows up everywhere at once, instead of hiding in one of three copies.
EByte is a Go library that works with any reader or writer, and one command-line tool with four commands: validate-schema, decode, encode and inspect. The same library also runs in the browser as WebAssembly, and the demo on the site is the real thing.
Schemas cover integers of 8 to 64 bits in either byte order, 32- and 64-bit floats, bit fields, ASCII and UTF-8 strings, raw byte fields, counted and nested arrays, blocks that repeat to the end of the file or while a signature matches, and equality conditions. Counts, sizes, offsets and checksums, CRC-32, Adler-32 and CRC-16, are computed by the encoder automatically. Edit a WAV file’s samples and the size fields update. Change a PNG chunk and its CRC is recomputed.
EByte can also compare two files field by field under one schema, which is handy for checking sensor captures before and after a firmware update, and it can import Kaitai Struct descriptions that fit what a schema can express.
Who It’s For
- Firmware and embedded engineers working with device logs, sensor captures and config blobs.
- Developers maintaining custom file formats who are tired of keeping reader, writer and docs in step.
- Reverse engineers and testers who want to edit a binary file safely and see exactly where it breaks.
- Game modders and archivists dealing with old or obscure formats.
Limits
There are hard limits of 16 MiB of input, 65,536 elements per array and 100,000 values, which you can lower but not raise. No compression, no reading backwards to an earlier offset, and no general arithmetic between fields. Only some Kaitai descriptions fit: 9 of the 191 in the Kaitai gallery. No code generation, no hosted service and nothing to sign up for, on purpose.
The Four-Question Screening
Every project featured here gets the same four questions, the way you’d screen a startup idea.
1. Where is the money in it?
EByte is free and has no paid tier. The money is in industries that live on binary formats: embedded devices, automotive and industrial telemetry, medical devices and test equipment. They pay for tools that cut debugging time and for schemas of their own formats written and maintained by someone who knows what they’re doing. Paid paths include a schema library for common industrial formats, a desktop editor built on the engine, and consulting to describe and validate proprietary formats.
2. What real problem does it solve?
Format descriptions that disagree with each other, and binary files that can’t be edited safely. The first causes bugs that show up only with certain files. The second makes simple fixes risky.
3. The technology in plain words
Think of a form with boxes: name here, age there, three lines for an address. EByte is the blank form. Give it a filled-in one written in a code you can’t read, and it copies the answers into plain words for you. Change an answer, and it writes the code back out, fixing any totals at the bottom so the form still adds up.
4. Moat and differentiation
Kaitai Struct is the best-known tool for describing binary formats, and it focuses on reading through generated code. EByte’s edge is one interpreter that reads, writes and traces from a single schema, with byte-for-byte round trips and no code generation. That makes it safe for editing, not just reading. The moat that can grow is the schema library and a reputation for exact round trips, which is what people working with real devices care about.
Why It Matters
A lot of the world’s data sits in binary formats nobody can read without a special tool. One clear description, used everywhere, makes those files inspectable and editable without fear of breaking them.
Follow the Project
Try it in the browser demo, read the plain-language explainer on binary schemas, start with the quick start, or get the code from the EByte repository on GitHub. EByte is listed with the other open-source projects on SoftDevStudio.com, and more featured projects are in the Projects section.