Claude Code · Statusline

Everything that matters, on one line.

See context usage, session limits, weekly allowance, Git status, and active tasks at a glance — without leaving Claude Code.

my-project ⎇ main ↑2↓1 Opus 4.8 (1M) · high C45 H14 ↺ 4h21m W31 ↺ 2d14h $44.21
$ npx ctxline-claude
Then restart Claude Code or start a new session. That's it. (bunx also works)
Works on macOS·Linux·Windows
What it shows

Nine signals, one line

Read left to right, it's the whole picture. Everything is best-effort and silent — if a source is missing, the line just skips it and still renders. On a narrow terminal it wraps to a second line; widen, and it returns to one.

↑ hover, tap, or use arrow keys to inspect a segment

DIR

Subagent panel

Subagents get their own line too

Keep an eye on delegated work without opening extra panels. Each running subagent gets a compact row showing its name, model, effort level, context usage, and elapsed time.


○ architecture-reviewOpus 5 · maxC51⏱ 4m12s
○ test-coverageSonnet 5 · highC36⏱ 1m48s

Under the hood

Fast, cached, never breaks

A single zero-dependency Node script. It always prints, and never hangs or throws.

01

Reads session

Context comes straight from Claude Code's session data on every render.

02

Reads usage

5-hour and weekly limits come from the session data — no network. It falls back to Anthropic's usage API when needed; credentials never leave your machine.

03

Caches it

The fallback fetch is cached and shared across sessions, with adaptive timeouts so the line stays snappy.

04

Falls back

Every failure path degrades silently to cached or partial output. The statusline always renders.

Learn more

The rest lives in the docs on GitHub — one click away.

Support

Enjoying ctxline?

Donations help keep it maintained, improve existing features, and fund new open-source tools.