statusline-pi

Compact project statusline footer for Pi.

Format:

current-dir │ branch [changed files] PR #x │ estimated session cost │ CPU % · MEM % │ remaining context tokens (percentage) context zone │ average response speed │ provider/model

Example:

pi-extensions │ main [2] PR #12 │ $0.18 │ CPU 42% · MEM 68% │ 😄 840,037 (84.0%) Plan │ 42.5 tps │ openai-codex/gpt-5.5

Context zone icons change with usage:

Behavior

• Installs as a Pi extension and enables automatically on session start.
• Replaces Pi's default footer with a compact responsive statusline.
• Uses one line when the terminal is wide enough, then wraps into multiple width-safe lines on narrow terminals so long branch names do not hide context, speed, or model details.
• Refreshes git change count and host CPU/memory usage every 5 seconds.
• Shows CPU and MEM utilization for the local machine (CPU 42% · MEM 68%). CPU is derived from os.cpus() time deltas (omitted until the second sample). Memory is (total - free) / total. Colors follow the same thresholds as other indicators: default success, warning at ≥85%, error at ≥95%.
• Shows average model response speed as output tokens per second (tps) across completed assistant responses.
• Shows a warning when an OpenAI GPT model reaches 272,000 context tokens, the reported pricing breakpoint where the same token costs double. The warning is emitted once per threshold crossing.
• Shows an estimated accumulated session cost in USD, summed from each assistant response's token usage (input, output, cache-read, cache-write) and the active model's per-million token rates from Pi's model catalog (aligned with pi.dev/models). This is an estimate only—actual billing may differ by provider, discounts, or OAuth subscriptions.
• Updates the cost after each assistant response and when you switch models; omits the cost segment when the active model has no pricing, and displays cost ? when usage was reported without a computable price.
• Includes the active assistant response in the average while it is streaming, then keeps the completed average visible while idle.
• Checks for a GitHub PR associated with the current branch every 60 seconds using gh pr view.
• Omits the PR segment when gh is unavailable or the branch has no PR.

Commands

• /statusline-pi — open a compact enable/disable menu in the TUI (Escape or Close leaves the current state unchanged); outside the TUI it keeps the original toggle behavior.
• /statusline-pi toggle|on|off — toggle or explicitly set the footer without opening a menu.
• /statusline-refresh — force refresh git and PR data (also available from the TUI menu).

Compatibility

The menu is shown only when Pi supplies an extension context with mode === "tui" and hasUI. Older host runtimes that omit ctx.mode use the non-TUI fallback instead of opening UI/RPC dialogs: the bare /statusline-pi command toggles the footer.

Settings persistence

The deliberate /statusline-pi toggle is stored in the private, validated ~/.pi/agent/statusline-pi.json file (the path follows Pi's getAgentDir()). The default remains enabled when no preference is present. Git, response-speed, cost, system, and other footer state stays session-only; refreshes do not write preferences.

Install

Published on npm: statusline-pi. Use Pi's package manager (pi install), not npm install alone.

pi install npm:statusline-pi
pi install npm:statusline-pi@1.3.1   # pin version
pi install -l npm:statusline-pi      # project-local (.pi/settings.json)
pi -e npm:statusline-pi              # one session, no install

Then run /reload in Pi (or restart).

pi list
pi update npm:statusline-pi
pi remove npm:statusline-pi

From pi-extensions (git):

cp -r extensions/statusline-pi ~/.pi/agent/extensions/