AI Agent Hub
Back to plugins
🖥️

dsh-usage-meter

Client Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install V-dev-388/dsh-usage-meter

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install V-dev-388/dsh-usage-meter in your terminal to install; the source repository is https://github.com/V-dev-388/dsh-usage-meter .

About this plugin

DSH users who have been running conversations for a while often lack a clear picture of total token consumption: which providers and models use the most, how well caching is working, and what the recent trend looks like. dsh-usage-meter adds a Usage section to the Settings page, aggregating token data from every session including sub-agent sessions into six summary metrics — requests, total tokens, input, output, cache read, and cache hit rate — plus per-provider and per-model breakdown tables.

Time-range views cover today as 24 hourly bars and the last 7 or 30 days as daily bars, all in local timezone, with 60-second auto-polling and a manual refresh button. Data is persisted locally under $DSH_HOME via atomic writes; on startup the plugin backfills historical transcripts through the official sessionQuery, live sessions are updated incrementally via events, and watermark-based deduplication ensures replay never double-counts. The plugin carries zero third-party dependencies, calling only official Cordis services and the built-in paths package, and is entirely passive with no network calls or telemetry.

Ideal for DSH users who want lightweight token accounting without pulling in external services or cost-estimation layers. The plugin reports raw token metrics and cache hit rate only — no fee conversion, no per-request log table — keeping the data semantics clean and predictable.

Use Cases

  • Glance at token consumption breakdown by provider and model in the Settings page
  • Compare usage trends across today, 7 days, and 30 days to spot peak hours
  • Track cache hit rate changes to evaluate prompt or model adjustments

Best For

  • DSH daily users who need a lightweight token usage overview
  • Developers monitoring multi-provider consumption without external SaaS
  • AI engineers tracking cache hit rates to optimize prompting strategies