@six6strings/dsh-client-ui-thinking-stats
一个轻量级插件，可在底部 Dock 和每次对话末尾添加模型思考令牌统计信息。
插件会安装到这里；不确定时保持 web。
npx -y @deepseek-ai/dsh plugin --profile web add github:Six6stRINgs/dsh-client-ui-thinking-stats#14654ed7e3c4477f80653eec245238527eefd17f
Client Ui Thinking Stats 以 @six6strings/dsh-client-ui-thinking-stats 发布，当前版本为 1.0.2。Plugin Hub 会校验它的 manifest，并保存精确安装来源，便于复现安装结果。
A lightweight plugin that adds model thinking token statistics to the bottom Dock and the end of each conversation. Client-only and zero-cost when idle: it reads the existing conversation snapshot and renders nothing at all when there is no thinking.
It adds two readouts, both of which render nothing when the model has produced no thinking for the relevant scope:
| Surface | Slot | Scope |
|---|---|---|
| Bottom composer dock | conversation.composer.dock | Whole session (cumulative) |
| Each assistant reply (timing row) | conversation.chat.assistant-actions | That single turn |
The per-turn readout is appended to the reply's timing strip — the same hover
row that shows 20:55 · 用时 11秒 · 首token 10秒 · 28 tok/s — so it reads as part
of that same unit.
Each readout shows, left to right, a brain glyph, the thinking-token count, its share of all tokens, and its share of output tokens, each with one decimal place, for example:
💭 15 · 0.2% · 60.0%
Hover for the exact numbers.
For every finalized assistant message, in order:
usage.reasoningTokens is present (DeepSeek,
OpenAI o-series, Anthropic, …), it is used as-is. This is exact.The two denominators are derived from the same assistant messages, so the percentages always agree with the count on one consistent scope:
lib/index.js) is an empty
apply that exists only so the package appears in the host Loader. All work is
done in the browser.null — no element, no layout cost.--dsw-alias-* tokens, so it follows
light/dark automatically.The plugin follows the standard dsh.bundle + dsh.client convention, so it
installs like any DSH plugin. From this repository:
dsh plugin add github:Six6stRINgs/dsh-client-ui-thinking-stats
Then restart dsh web and reload the page. The readouts appear only once the
model starts thinking.
test/harness.mjs stubs the browser/React environment, runs the real factory
and apply, and renders both entries against sample data (provider-reported and
block-estimated paths, the one-decimal formatting, and the hidden-when-empty
cases). Run with:
node test/harness.mjs
reasoningTokens is always preferred.