@tian.zuo/pi-token-speed

Live and final tokens-per-second meter for the pi coding agent.

Packages

Package details

extension

Install @tian.zuo/pi-token-speed from npm and Pi will load the resources declared by the package manifest.

npm repo home report

$ pi install npm:@tian.zuo/pi-token-speed

Package: @tian.zuo/pi-token-speed
Version: 0.1.1
Published: Sep 22, 2026
Downloads: 237/mo · 14/wk
Author: tian.zuo
License: MIT
Types: extension
Size: 17.9 KB
Dependencies: 1 dependency · 1 peer

Pi manifest JSON

{
  "extensions": [
    "./index.ts"
  ]
}

Security note

Pi packages can execute code and influence agent behavior. Review the source before installing third-party packages.

README

@tian.zuo/pi-token-speed

Release notes: changelog · GitHub releases

Live and final tokens-per-second reporting for the pi coding agent.

pi install npm:@tian.zuo/pi-token-speed

How it works

While the assistant streams, the footer shows a smoothed tokens-per-second rate; when the message finishes it shows a summary with the average rate, total output tokens, and time-to-first-token. The summary stays on screen after the stream stops — including the model's between-stream thinking and tool-call gaps — so the readout is always visible instead of blanking out whenever generation pauses.

The live rate is sampled from streamed text (a responsive chars-per-token estimate); the end-of-message average uses the provider's authoritative output token count when available. The mode is remembered in ~/.pi/token-speed/config.json.

Commands

/tps — cycle the display mode: live → final → off.
/tps live — live meter + summary.
/tps final — summary only.
/tps off — show nothing.

@tian.zuo/pi-token-speed

How it works

Commands

License