Plugin directory / Knowledge / dsh-bottom-stats
dsh-bottom-stats
Verified · install-tested on dsh 318197375
What it does
DSH plugin: full-width conversation stats line (no truncation) + context occupancy progress bar for the DeepSeek Harness web UI
Works — verified, early-stage project
DSH plugin: full-width conversation stats line (no truncation) + context occupancy progress bar for the DeepSeek Harness web UI It installs cleanly and boots without issues in our testing. It's early-stage but functional.
“Verified” means our automated CI actually ran dsh plugin add in a clean profile and it booted — nothing more. Feature descriptions and version compatibility are the author’s claims. This is not a security audit and not an endorsement of third-party code.
README
dsh-bottom-stats
DeepSeek Harness(DSH)Web UI 插件 · 底部统计栏增强
DeepSeek Harness (DSH) web UI plugin — enhanced bottom stats strip

中文
这是什么
为 DSH Web 对话页输入框下方提供两行常驻信息:一条不再被截断的统计行,和一条上下文占用进度条。
功能
- 统计行:轮/步 · LLM/工具耗时 · 首 token/吞吐 · 缓存命中 · 输入/输出 token。整行占满主栏宽度并居中——对话再长、输入 token 涨到数百万,末尾的“输出 … tok”也始终完整可见(官方统计行超长时会被省略号截断)。
- 上下文占用进度条:限定在对话内容列内居中,按 系统提示词 / 工具 / 对话消息 三色分段显示占比,右侧显示占用百分比与
已用/总量token(如18% 177K/1M);悬停进度条可查看完整读数与各项占比明细。 - 数据与官方 UI 完全同源(
sessionStats、tokenUsage、contextPressure、contextBreakdown投影),文案跟随系统语言(中文/English)。
安装
npx -p @deepseek-ai/dsh dsh plugin --profile web add github:318197375/dsh-bottom-stats
安装后重启 web profile(bundle 层变化需要重启,刷新页面不够)。
纯 JS 分发,无
prepare构建脚本,不需要allowBuilds授权。
使用
安装并重启后即可生效:任意会话的输入框下方会自动出现上述两行,无需额外操作。
卸载
dsh plugin --profile web remove dsh-bottom-stats
兼容性
- 需要 DSH Web UI(
platform: web)。 - 依赖官方
conversation.composer.dock槽位与sessionStats/tokenUsage/contextPressure/contextBreakdown投影,标准 Web 组合均已提供。
许可
MIT
English
What is this
Adds two persistent rows beneath the composer in the DSH web chat: a stats line that never truncates, and a context-occupancy progress bar.
Features
- Stats line: turns/steps · LLM/tool time · TTFT/throughput · cache hit · input/output tokens. The whole line spans and centers on the main column — no matter how long the conversation grows (input tokens in the millions), the trailing
output … tokgroup stays fully visible (the stock line elides it with an ellipsis when it overflows). - Context-occupancy bar: centered within the chat content column, segmented by system prompt / tools / messages proportions, with the occupancy percent and
used/totaltokens on the right (e.g.18% 177K/1M); hover the bar for the full reading and per-part breakdown. - All data comes from the same projections the stock UI uses (
sessionStats,tokenUsage,contextPressure,contextBreakdown); labels follow the UI language (Chinese/English).
Install
npx -p @deepseek-ai/dsh dsh plugin --profile web add github:318197375/dsh-bottom-stats
Restart the web profile after installing (bundle-layer changes need a restart; refreshing the page is not enough).
Pure-JS distribution — no
preparescript, so noallowBuildsauthorization is required.
Usage
Nothing to configure: after install + restart, the two rows appear under the composer of any active conversation.
Uninstall
dsh plugin --profile web remove dsh-bottom-stats
Requirements
- DSH web UI (
platform: web). - Uses the stock
conversation.composer.dockslot and thesessionStats/tokenUsage/contextPressure/contextBreakdownprojections, all present in standard web compositions.
License
MIT
Install
Install the catalog once, then DeepSeek Harness can find and install any plugin from this site automatically:
dsh plugin add dshbase-catalog Then say "install dsh-bottom-stats for me" — your agent finds it in the directory and installs it. Docs: dshbase-catalog · verified packs.
This plugin is GitHub source (not published to npm) — install it straight from the repo:
Web profile:
dsh plugin --profile web add github:318197375/dsh-bottom-stats Headless (CLI) profile:
dsh plugin --profile headless add github:318197375/dsh-bottom-stats Test report
Verified: L1 install + L2 load + L3 runtime from GitHub source on dsh 0.1.0-rc.6.
When to use it
Give the agent a memory, a knowledge base, or a retrieval layer so it stops forgetting context between sessions.
Who it's for
Users running long projects who want the agent to remember decisions, docs, and preferences without re-explaining.
For developers — extending it
The memory/retrieval backend is the seam — plug a new store, tune what gets distilled, or add citation and audit trails.