Status line

Inside a Claude Code session your shell prompt is hidden, which is exactly when knowing the account matters most. The status line puts it back.

---

claude-profiles statusline

Inside a Claude Code session the shell prompt is gone, so the profile indicator is not visible. This renders a status line for Claude Code's own statusLine setting instead.

work │ you@company.com │ ⎇ main │ ctx 52%/1000k │ 5h 2% │ 7d 3% │ $12.40

Enabling it

claude-profiles statusline --install                  # the active profile
claude-profiles statusline --install-all              # every profile
claude-profiles statusline --install-all --show profile,limits   # pick the segments

That writes the setting into each profile's settings.json, backing up any existing file alongside. Settings are per-profile, which is why --install-all exists. An existing status line from somewhere else is left alone unless you pass --force.

Start a new Claude Code session to see it - the setting is read at launch.

Or configure it by hand

{
  "statusLine": {
    "type": "command",
    "command": "python3 \"$HOME/.claude-tools/bin/claude-profiles.py\" statusline --show profile,account,limits",
    "padding": 0
  }
}

Segments

--show takes a comma-separated list, rendered in the order you give:

Segment	Shows
profile	the active profile - dim on default, magenta otherwise
account	the signed-in email
model	model name, plus effort level when it is not the default
dir	the current directory's basename
branch	the git branch
context	context window used, colour-coded
limits	5-hour and 7-day usage, each its own segment
cost	what this session's tokens would cost at API rates
lines	lines added and removed this session
version	the Claude Code version
session	the session name

Percentages are green below 75%, amber from 75%, red from 90% - the same scale claude-profiles uses.

cost is not a bill

It is total_cost_usd from the payload: what this session's tokens would cost at pay-as-you-go API rates. On a subscription you are not charged it - treat it as a sense of how heavy the session is. Leave the segment out if that number is more distracting than useful.

These figures come from Claude Code, not the cache

Claude Code passes current rate limits and context usage to the status line on every render, so limits here is live without any network call of its own. If a payload ever lacks them, it falls back to the cached snapshot.

Check for and pull a newer version. Covered in full on the Updating page.

claude-update --check    # what is installed, and what is available
claude-update            # pull it

Flag	Effect
-c, --check	report only, change nothing

It refuses to run over uncommitted local edits, uses git pull --ff-only, and tells you whether the change needs a shell restart.

---