DeepSeek Harness Plugin Hub

发布与管理完整 Harness Profiles,发现适合你的插件。

探索

插件目录环境预设文档中心动态

社区

发布插件联系我们报告问题

相关链接

Plugin Hub GitHubDeepSeek Harness 官方项目系统状态隐私说明
© 2026 DeepSeek Harness Plugin HubPowered byPaxTech

独立、非官方社区项目,与 DeepSeek 官方无隶属、授权或背书关系。

DeepSeek Harness Plugin Hub
ProfilesPlugins分类动态文档登录管理 Profiles
ProfilesPlugins分类动态文档登录
← Plugins
T

@apodemakeles/dsh-token-dashboard

Token Dashboard

DSH 网页 GUI 令牌消耗热力图:由 session-log 使用事件提供数据的每日/每周令牌总使用量面板

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:apodemakeles/dsh-token-dashboard#6b45910d29927402387529b5250c764c1c28f7d9
README兼容性版本
Weekly token-usage heatmapDaily token-usage bar chart

兼容性与来源证明

Token Dashboard 以 @apodemakeles/dsh-token-dashboard 发布,当前版本为 0.2.1。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。

DSH 兼容范围
*
运行环境
web
发布来源
github
Registry 更新时间
2026/10/26

版本

0.2.1stable
2026/10/26
0.2.0stable
2026/10/20

相关插件

正在加载相关插件…

最新版
0.2.1
DSH
*
HMR
重启进程
Tree shaking
未声明可安全裁剪
解包体积
未提供
文件数
未提供
Surface
web
许可证
MIT
发布源
github
GitHub
★ 1
周下载
0
最近提交
2026/10/26
查看源码 ↗
README Badge

点击下方 Badge 复制 Markdown,粘贴到 README 即可。

这是你的 Plugin?认领权益 · 优先安全扫描

验证 package.json 声明的 GitHub 仓库,即可管理这个公开页面。认领后,Hub 会优先安排当前版本的安全扫描,并在通过后公开展示结果。

认领这个 Plugin →
报告问题

相关插件

继续浏览 models-usage 分类下经过校验的插件。

Experimental Cot Translation Bundle@deepseek-ai/dsh-experimental-cot-translation-bundle扩展推理的实验性机器翻译Our Free Model@ebony-vinyl/dsh-our-free-modelOur Free Model — 在 dsh 里装上这个插件即可,无需登录、注册或填 API Key,就能使用包括 DeepSeek V4.1 Flash、Kimi K3 在内的前沿模型,桌面端另有 EAC 协付渠道;2.0 起接入 LobsterAI、Qoder、TRAE、Cline、Loomy、Raccoon、MiniMax、ZCode、Gemini 十个账号渠道(原有匿名免费通道保留)(登录、账号池、积分领取、本地 OpenAI 网关与局域网转发、数据看板与运行日志),六页毛玻璃界面统一呈现。Usage@linxin666/dsh-usage用于 dsh Web GUI 的使用统计插件:检测各提供商的余额和编程套餐配额,并实时记录令牌使用情况;侧边栏 Settings 行上方显示紧凑的价格概览卡片Client Ui Model Capabilities@linxin666/dsh-client-ui-model-capabilities为自定义 pi-ai 提供商声明每个模型的能力(图像输入和推理力度),可直接在 Models 设置卡片中编辑;写入官方 llm-pi-ai 设置命名空间,不更改 DSH 源代码。

README

@apodemakeles/dsh-token-dashboard

English | 中文

A DeepSeek Harness (DSH) web GUI plugin: a token-consumption heatmap panel. It shows your total token usage per day / per week across all projects on this machine, GitHub-contributions style. Usage facts are projected into a local SQLite database by a persistent Worker, so opening the panel never scans session logs.

Status: released v0.2.0 — GitHub distribution is live. The package is not currently published to npm.

Screenshots

Weekly heatmap

Weekly token-usage heatmap

Daily view

Daily token-usage bar chart

What it does

  • Host half — listens to live session/event, flushes through DSH's durability barrier, and sends minimal usage deltas to a persistent Worker. The Worker owns SQLite (node:sqlite), commits facts/checkpoints atomically, and serves one consistent snapshot endpoint.
  • Client half — puts a live today capsule in the sidebar footer (● 今日 478.9k tokens): today's total, always visible without opening anything, refreshed by one cheap weeks=1 snapshot read on mount, on window focus, when the panel closes, and every 30s while the window is visible. Clicking it opens the panel — weekly heatmap grid + daily view — which fetches a single GET /api/token-dashboard/snapshot per open/refresh/page change.
  • CLI — dsh-token-dashboard status|verify|rebuild|backups|restore|cleanup for local maintenance.

Installation

The plugin is one package with two halves and no runtime dependencies; it runs on the DSH web profile and in the official DeepSeek Harness desktop app (which boots the same web App inside Electron on its own desktop profile).

Desktop app

The CLI refuses --profile desktop ("managed exclusively by the Electron application"), so install into the desktop profile the way the app's own plugin manager does — declare the dependency and run pnpm in the profile directory:

DSH_HOME=${DSH_HOME:-$HOME/.dsh}
PROFILE="$DSH_HOME/profiles/desktop"

# 1. add the dependency to the desktop profile manifest and re-run pnpm
pnpm --dir "$PROFILE" add github:apodemakeles/dsh-token-dashboard   # or: link:/path/to/checkout

Then add the package to that manifest's dsh.profile.bundles list, so the profile loads its cordis.patch.yml layer:

{ "dsh": { "profile": { "bundles": [
  "@deepseek-ai/dsh-base",
  "@deepseek-ai/dsh-web-app",
  "@apodemakeles/dsh-token-dashboard"
] } } }

Restart the desktop app. The app's 插件 → 已安装 screen lists the package as an enabled plugin, and the 今日 N tokens capsule appears in the sidebar footer.

Web profile

From GitHub:

dsh plugin --profile web add github:apodemakeles/dsh-token-dashboard

Then restart dsh web. The sidebar footer shows the 今日 N tokens capsule; click it to open the heatmap panel.

The npm package is not currently published; use the GitHub installation above.

Data and operations

  • Projection database: $DSH_HOME/data/token-dashboard/usage-v1.sqlite (default ~/.dsh).
  • Profiles share one Harness home, so the desktop app and dsh web read the same database and the same $DSH_HOME/sessions logs: history collected under one profile shows up under the other with no migration step.
  • Total tokens = inputTokens + outputTokens + cacheReadTokens (cache reads count into the headline).
  • The panel only calls the snapshot route; it does not trigger a session listing, inspect, a source read, flush, or backfill.
  • First startup initializes the database in the background; the panel shows phase/progress until ready.
  • A session log the running harness cannot interpret (for example a pre-release subagent log whose child descriptors are no longer readable) is isolated: it is listed on stderr with its reason, the phase ends degraded instead of ready, and the facts already committed for it stay in the database.

Development

git clone https://github.com/apodemakeles/dsh-token-dashboard.git
cd dsh-token-dashboard
pnpm install && pnpm build
dsh plugin --profile web add link:$(pwd)

For the desktop app, point it at the checkout instead:

pnpm --dir "${DSH_HOME:-$HOME/.dsh}/profiles/desktop" add link:$(pwd)

See .scratch/dsh-token-dashboard/map.md for the project map (planning tracker), docs/durable-usage-architecture.md for the architecture contract, and docs/dev-loop.md for the full dev loop.

Known limitations

  • Projection is machine-local: it reads sessions stored on this machine (~/.dsh/sessions is the authority; SQLite is a rebuildable projection).
  • Host/Worker changes require restarting DSH; client HMR does not cover the persistent Worker.
  • SQLite is experimental in Node 24 (node:sqlite); the plugin checks capability at startup and does not hide the warning globally.

License

MIT