Terminal · intermediate

Splitting output with tee

tee copies bytes from standard input to standard output and simultaneously writes copies to one or more files.

Why it matters

It preserves a live pipeline while recording or observing intermediate data that would otherwise require rerunning an expensive producer.

Mental model

How to reason about splitting output with tee

For each block read from stdin, write the same bytes to every named file and to stdout; downstream processing continues from that stdout copy.

It is a plumbing T-junction that sends the same flow onward while filling a side reservoir for inspection or storage.

Examples

Worked example · Record an intermediate stream

curl -fsS "$URL" | tee response.json | jq '.items | length'

The exact response bytes are saved to response.json and also continue to jq, avoiding a second network request.

Worked example · Append to a log

run_checks 2>&1 | tee -a checks.log

The shell combines stderr with stdout before the pipe, and tee -a appends the combined stream while still displaying it.

Useful contrast · Redirection ends the branch

generate > output.txt

Plain redirection sends stdout only to the file; tee is needed when the same bytes must also remain available on stdout.

Common mistakes

Use -a only when appending is intended; otherwise each named output file is opened for overwrite.

A pipe receives stdout by default, so stderr must be deliberately combined or separately handled before tee can copy it.

Quick check

1. What two destinations does tee output.txt write to by default?

Answer: output.txt and standard output

2. Why does command | tee log not normally capture command's stderr?

Answer: The pipe connects only command's stdout unless stderr is explicitly redirected into it.

Keep building