Output formats

Every command streams its records through one renderer, so the output flags work the same everywhere.

Choosing a format

-o / --output takes:

Format	Use
auto	table on a terminal, JSONL when piped (the default)
table	aligned columns for reading
json	a single JSON array
jsonl	one JSON object per line, the streaming format
csv	comma-separated, with a header
tsv	tab-separated, with a header
url	just the URL of each record, one per line
raw	the underlying HTML/JSON amz fetched

amz product B084DWG2VQ -o json
amz search "usb c cable" -o jsonl
amz bestsellers electronics -o csv > top.csv

Because auto switches on whether stdout is a terminal, the same command reads nicely by hand and pipes cleanly in a script, with nothing to remember.

Projecting fields

--fields picks and orders columns by their JSON name. It applies to every format, so it trims a CSV as readily as a table:

amz product B084DWG2VQ -o csv --fields asin,price,rating
amz search "usb c cable" --fields asin,title,price -o table

--no-header drops the header row from table, CSV, and TSV.

Templates

--template renders each record through a Go text/template, for a custom line format:

amz search "usb c cable" --template '{{.asin}}  {{.price}}  {{.title}}'

Writing to a file

-O / --out writes the rendered output to a file instead of stdout (and forces non-TTY formatting):

amz reviews B084DWG2VQ -o csv -O reviews.csv

Exit codes

amz uses its exit code to tell apart the ways a command can end, so a script can branch without parsing output:

Code	Meaning
0	success
1	runtime error
2	usage error (bad flag, unknown marketplace)
3	no data (the surface was empty)
4	partial (some pages fetched, some failed)
5	blocked (Amazon served the bot wall)

See troubleshooting for what to do with code 5.