RecordStreamYour data's new best friend

See It in Action ​

Got a CSV? Turn it into insights in three lines:

# Who's spending the most? Let's find out.
recs fromcsv --header purchases.csv          \
  | recs collate --key customer -a sum,amount \
  | recs sort --key amount=-n                 \
  | recs totable

customer   sum_amount
--------   ----------
Alice      9402.50
Bob        7281.00
Charlie    3104.75

Or maybe you've got some JSON and you need answers now:

cat api-response.json                       \
  | recs fromjsonarray                      \
  | recs grep '{{status}} === "active"'     \
  | recs xform '{{age}} = {{age}} + 1'      \
  | recs totable -k name,status,age

Install ​

# One-liner (macOS / Linux)
curl -fsSL https://raw.githubusercontent.com/benbernard/RecordStream/master/install.sh | bash

That detects your platform, downloads the right binary, and puts it in your $PATH. Updates happen automatically — recs checks for new versions in the background and tells you when one is available. Run recs --update to upgrade in place.

Philosophy ​

RecordStream is built on a simple idea: one JSON object per line. Every command reads records from stdin, does something useful, and writes records to stdout. Chain them with pipes and you've got a data pipeline that would make a shell wizard weep with joy.

Input commands create records from the outside world — CSV files, databases, XML, Apache logs, you name it.

Transform commands reshape, filter, sort, collate, and generally boss your data around.

Output commands turn records into something humans can read — tables, CSV, HTML, pretty-printed JSON.

The result? Complex data transformations expressed as readable, composable, debuggable pipelines. No Spark cluster required.