Output and pipelines

Every read in x produces rows. The same rows can come out as a table, JSONL, JSON, CSV, TSV, plain URLs, raw bytes, or whatever a Go template makes of them. This guide shows how to pick a shape and pipe it.

Formats

-o (or --output) selects the format:

x timeline nasa -o table     # aligned columns (default on a terminal)
x timeline nasa -o jsonl     # one JSON object per line (default when piped)
x timeline nasa -o json      # a single JSON array
x timeline nasa -o csv       # comma-separated, with a header row
x timeline nasa -o tsv       # tab-separated
x timeline nasa -o url       # just the URL of each row
x timeline nasa -o raw       # the raw upstream payload, unshaped

When you do not pass -o, x chooses for you: table on a terminal, jsonl when the output is a pipe or file. So x timeline nasa reads nicely on screen and x timeline nasa | program feeds JSONL without a flag. Drop the header row from csv/tsv with --no-header.

Projecting columns

--fields keeps only the columns you name, in order:

x timeline nasa --fields id,text
x following nasa --guest --fields username,name -o csv

The field names are the JSON keys of a row (see the template section for how to discover them).

Templates

--template renders each row with Go's text/template. The row is the dot, and its fields are the JSON-tag keys, including nested ones:

x timeline nasa --template '{{.id}} {{.author.username}}: {{.text}}'
x user nasa     --template '{{.username}} has {{.metrics.followers}} followers'

Keys mirror the JSONL output: top-level fields like {{.id}} and {{.text}}, nested objects like {{.author.username}}, and metric counts like {{.metrics.followers}}. Integers render as plain digits, so a template is a clean way to build a custom line per tweet or account. To see the available keys for a command, run it once with -o json and read the structure.

Piping to jq

JSONL is the natural bridge to jq:

x search "from:nasa" --guest -o jsonl | jq -r .id
x followers nasa --guest -o jsonl | jq -r 'select(.metrics.followers > 1000) | .username'

IDs are strings

X IDs are 64-bit snowflakes. If a tool parses them as numbers it silently corrupts the low digits. x always emits IDs as strings, so they survive jq, spreadsheets, and round-trips untouched:

x tweet 20 -o json | jq .id     # "20", a string, not 20

A 19-digit tweet id comes back exactly as sent. Build URLs and re-query with the string as-is.

Limiting rows

-n (or --limit) caps the number of rows; 0 means unlimited:

x timeline nasa --guest -n 100   # at most 100 tweets
x followers nasa --guest -n 0    # everything the tier will return

x pages under the hood until it has enough rows or the source runs out, honoring the request rate between pages so it stays a polite client.