3,798 stars · MIT, read from /blob/master/LICENSE-MIT on 2026-09-29; it carries TWO copyright lines, "(c) 2015 Andrew Gallant" and "(c) 2026 datHere, Inc.", which is the xsv lineage the README never states · 23.0.1 (2026-09-13), the "Context is King" release, 270 commits since 22.0.1. Code on 2026-09-28. · Track this in Scout
A fast command-line toolkit for sorting, joining, checking and describing CSV files, with around 140 commands.
▶Repo detailsthe review · specs · pros & cons · install
What it is
A Rust program at 3,798 stars with around 140 commands and subcommands, at version 23.0.1 since 13 September 2026 and taking code the day before this edition. Its licence file is the detail worth reading: it carries two copyright lines, "© 2015 Andrew Gallant" and "© 2026 datHere, Inc."
What it is good for. Anyone who receives data as CSV files from someone else, which is most people who receive data at all. The commands that pay for themselves first are stats, which describes every column in one pass, and validate, which checks a file against rules before you load it anywhere.
- Very fast, and most commands stream, so a file larger than the machine's memory is still workable.
statson an unfamiliar file immediately answers the questions you were about to ask by hand: types, ranges, empties, cardinality.- MIT licensed, read from
LICENSE-MITon themasterbranch, plain and unconditional.
- Some commands load the whole file into memory, and the documentation marks them. On a very large file you need the streaming alternatives, and knowing which is which is on you.
- There are three prebuilt binaries with different feature sets, called
qsv,qsvliteandqsvdp, plus variants built with a different Rust compiler. You have to choose before you download. - The same company sells a paid graphical product called qsv pro. The command-line tool is genuinely MIT and complete, but it is one half of a commercial offering, and that shapes where effort goes.
- BurntSushi/xsv
10,800 stars, Unlicense and MIT, and archived by its owner on 24 April 2025; its README now recommends qsv or xan in its place, which is the author of the original naming its successors.
Track this in Scout 
medialab/xan4,500 stars, the other project named in that sentence; it leans towards exploring and drawing data in the terminal, where qsv leans towards processing pipelines.
Track this in Scout
johnkerl/miller10,000 stars, BSD-2-Clause; it handles CSV, TSV and line-by-line JSON with a small query language of its own, which is more expressive and slower to learn.
Track this in Scout
https://github.com/dathere/qsv/releases/latest
