jq: JSON Processing

Why JSON broke the line-oriented pipe

Pipes assume one record per line. JSON is nested. grep, cut, and awk can't follow a key into a sub-object. jq exists to make JSON pipeable — read from stdin, query with a small expression language, write JSON or text to stdout.

Identity and pretty-printing

jq '.' file.json = pretty-print and color (the identity filter). cat data.json | jq is the canonical "make this readable" command.

Selecting fields

• jq '.name' — value of name at the top level.
• jq '.users[0].email' — first user's email.
• jq '.users[].email' — every user's email (one per line).
• jq '.users | length' — count.
• jq '.users | map(.email)' — array of emails.

Filtering

• jq '.users[] | select(.active == true)' — active users.
• jq '.users[] | select(.age > 30) | .name' — names of users over 30.

Reshaping output

• jq '{id: .id, name: .name}' — new object with subset.
• jq -r '.[].url' — raw output, no quotes (great for piping URLs into curl).
• jq -c '.' — compact, one JSON object per line (newline-delimited JSON, the streaming-friendly format).

Real-world combos

curl -s api.github.com/repos/cli/cli/issues \
  | jq '.[] | {n: .number, title}' | head
kubectl get pods -o json \
  | jq '.items[] | select(.status.phase != "Running") | .metadata.name'

Both feed JSON from one tool into a structured filter and back to a human-readable result. That's pipes in 2026.