4531 stars · Unlicense OR MIT — read from /blob/master/UNLICENSE and /blob/master/LICENSE-MIT, both plain and unmodified; nothing at LICENSE, so automatic licence checkers find nothing · 0.61.0 of 2026-09-11, the yearless GitHub date settled to the minute by crates.io's API · Track this in Scout
A command-line tool for processing, exploring and drawing CSV data in the terminal.
▶Repo detailsthe review · specs · pros & cons · install
What it is
A program written in Rust, at 4,531 stars, maintained by a social-science research lab in Paris. It began as a fork of BurntSushi/xsv, the original fast CSV toolkit, and its README says it has been nearly entirely rewritten since. Its newest release, 0.61.0, is dated 11 September 2026, and code landed on 29 September 2026.
What it is good for. Anyone who receives a table too large for a spreadsheet and needs to answer one question about it: how many rows, which values appear, what the totals are per group. It is also useful as the first step before a table goes into a database, because it can describe and check a file before anything loads it.
- The lineage is credited honestly, which is rarer than it should be. The README says it began as a fork of xsv, and the licence file still names xsv's original author beside the current maintainer.
- Both licences are as permissive as they come: the project is offered as public domain or MIT, at your choice.
- Real speed. It parses using processor instructions designed for this kind of work and spreads the job across processor cores, so a large file is a wait of seconds rather than minutes.
- It has no version 1.0. At 0.61.0, command names and options can still move between versions, so a script written today is not guaranteed to run next year.
- There are 120 open issues against a single open pull request, from a small academic lab. That is a real risk if the lab's attention moves elsewhere.
- Operations such as sorting, joining and removing duplicates have to hold data in memory. "Handles gigabyte files" is not the same as "never runs out of memory", and no figures are published either way.
- There is no file at the usual
LICENSEpath: the two licences sit inUNLICENSEandLICENSE-MIT. Automatic licence checkers report nothing at all, which causes arguments in companies that run them.
dathere/qsvThe other successor named in xsv's own farewell note, doing the same job with a much larger command set, but its README does not credit xsv or its author where xan credits both openly.
Track this in Scout- johnkerl/miller
The same work on CSV, tab-separated files and line-by-line JSON with genuine streaming for files larger than memory, but written in Go and with a query language of its own to learn.
Track this in Scout - BurntSushi/xsv
The original both successors descend from, with the same commands and the same job, archived by its owner on 24 April 2025 and now pointing readers at the two projects that replaced it.
Track this in Scout
brew install xan sudo pacman -S xan scoop install xan
