Skip to content

1HelloMan1/dsh-stats-dashboard

40Last commit Aug 14, 2026

dsh-stats-dashboard DSH plugin

This plugin adds a fully interactive statistics dashboard to the DSH web interface. It aggregates session data to show key metrics like cache hit rate, response latency, token consumption, and estimated costs based on built-in DeepSeek pricing. Users can filter by provider, model, session, and date range, and export filtered logs to CSV.

How to install the dsh-stats-dashboard DSH plugin

dsh plugin --profile web add "github:1HelloMan1/dsh-stats-dashboard#main

Copying does not run this command. Review the repository and version before installing the dsh-stats-dashboard DSH plugin.

dsh-stats-dashboard DSH plugin data source

dsh-stats-dashboard DSH plugin snapshot date: Aug 16, 2026

discovered

What the dsh-stats-dashboard DSH plugin can do

  • Displays 7 core metric cards: total calls, total cost, cache rate, output tokens, input tokens, cache read tokens, cache write tokens.
  • Provides a detailed model statistics table with 12 columns including average latency, TTFT, throughput, and cost per model.
  • Supports real-time data refresh, multi-dimensional filtering (provider, model, session, date range), and CSV export with UTF-8 BOM encoding.
  • Includes a built-in pricing table for DeepSeek official models (deepseek-v4-flash, deepseek-v4-pro, etc.) and allows custom pricing via code modification.
  • Shows the latest 500 call logs with full performance and cost details.

Where the dsh-stats-dashboard DSH plugin fits

  • Monitor daily API usage and cost across different DeepSeek models and providers.
  • Analyze cache hit rates to optimize prompt caching strategies.
  • Track response speed (TTFT, throughput) per model to identify performance bottlenecks.
  • Export filtered call logs for auditing or billing reconciliation.
  • Quickly compare costs between different providers (DeepSeek, OpenCode, etc.) in a single view.

Who the dsh-stats-dashboard DSH plugin is for

  • DSH users who need to visualize and analyze their AI model usage costs and performance.
  • Developers managing multiple providers or models and requiring granular usage statistics.

dsh-stats-dashboard DSH plugin limitations

  • Only works within the DSH web interface; requires a running DSH profile with web mode.
  • Pricing data is based on a built-in table; adding custom models or providers requires editing the source code.
  • Currently installable via GitHub source; npm package is not yet published.
  • Call logs are limited to the most recent 500 entries.

dsh-stats-dashboard DSH plugin: from the repository README

Quoted from the 1HelloMan1/dsh-stats-dashboard README, the upstream source of the dsh-stats-dashboard DSH plugin. Copyright remains with the original authors.

> 📊 **DeepSeek Harness 插件** — 用量统计看板:按供应商/模型汇总响应速度、调用日志、Token 用量与费用估算 --- ## ✨ 功能特点 ### 核心指标 - **缓存命中率** — 全局和分模型的缓存效率一目了然 - **响应速度** — 平均耗时、首 Token 延迟 (TTFT)、吞吐量 (tok/s) - **Token 用量** — 输入/输出/缓存读/缓存写 完整统计 - **费用估算** — 基于内置 DeepSeek 官方定价表实时计算 ### 多维度分析 - **供应商维度** — 支持 DeepSeek、OpenCode、OpenCode-Go 等多供应商 - **模型维度** — 按 deepseek-v4-flash、deepseek-v4-pro 等分模型统计 - **会话维度** — 跨会话汇总,单会话筛选 - **时间维度** — 日期范围筛选,实时更新 ### 交互功能 - 🔄 **实时刷新** — 手动刷新获取最新数据 - 📥 **CSV 导出** — 导出筛选后的调用日志 - 🔍 **多维筛选** — 搜索/供应商/模型/会话/日期范围 - ⬆️⬇️ **表格排序** — 点击列头按任意指标排序 ### 数据展示 - **7 张总览卡** — 调用次数、费用、缓存率、输出、输入、缓存读、缓存写 - **供应商芯片** — 快速查看各供应商用量和费用 - **模型统计表** — 12 列详细指标,按需排序 - **调用日志** — 最近 500 条调用明细,含完整费用和性能指标 --- ## 📦 安装 ### 方式一:通过 DSH 插件命令(推荐) ```bash dsh plugin --profile web add "github:1HelloMan1/dsh-stats-dashboard#main" ``` ### 方式二:手动安装 ```bash # 克隆仓库 git clone https://github.com/1HelloMan1/dsh-stats-dashboard.git # 复制到 DSH profile 的 node_modules cp -r dsh-stats-dashboard ~/.dsh/profiles/node_modules/ # 编辑 DSH profile 的 cordis.patch.yml,添加插件 insert echo ' - insert: - id: stats-dashboard name: "dsh-stats-dashboard" ' >> ~/.dsh/profiles/web/cordis.patch.yml # 重启 DSH kill $(pgrep -f "ds

Read the full READMEThe repository declares no license. Check with the authors before using it.

dsh-stats-dashboard DSH plugin questions

How do I install the DSH stats dashboard plugin?

The recommended way is to run `dsh plugin --profile web add "github:1HelloMan1/dsh-stats-dashboard#main"` in your terminal. This will install the plugin directly from GitHub. You can also manually clone the repo and copy it to your DSH profile's node_modules directory, then add an entry to cordis.patch.yml.

How can I export the usage data to a CSV file?

First, apply any filters you want (provider, model, date range, etc.) on the dashboard. Then click the 'Export CSV' button. The file will be downloaded as `dsh-stats-YYYY-MM-DD.csv` with UTF-8 BOM encoding, so it opens correctly in Excel without garbled characters.

Can I add custom pricing for models not in the built-in table?

Yes, you can edit the `PRICING` object in `lib/index.js` of the plugin. The format is `"provider\0model": { inputUncachedCny, cacheReadCny, outputCny }`. After modifying, refresh the DSH page to see the updated cost estimates. Note that this requires manual code changes.

Does the dashboard update automatically?

The dashboard shows data from your DSH session logs. It does not auto-refresh in real time; you need to click the refresh button to get the latest data. The data is aggregated from the `session.list` RPC, so it reflects the state of your sessions at the moment of refresh.

Why can I only see the last 500 call logs?

The plugin currently limits the displayed call log to the most recent 500 entries for performance reasons. The summary statistics (total calls, cost, etc.) are still calculated across all sessions, but the detailed log table shows only the latest 500 calls.