AI Agent Hub
Back to plugins
🖥️

dsh-token-meter-mood

Client Updated 2026.08.23

Run the following command in DeepSeek Harness:

dsh plugin install firekula/dsh-token-meter-mood

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

Run dsh plugin install firekula/dsh-token-meter-mood (source: https://github.com/firekula/dsh-token-meter-mood) in a terminal where the Web profile is already initialized, then restart dsh --profile web or refresh the browser to see the per-turn token breakdown and mood emoji beneath each assistant reply.

About this plugin

When you work through multi-turn conversations in the DeepSeek Harness Web GUI, per-round token costs are buried deep inside session logs. Pulling them out means digging through JSON and summing numbers by hand, which is tedious and gives you little intuitive sense of what you actually consumed. dsh-token-meter-mood pins a compact usage strip directly beneath every assistant reply, showing input, cache-hit, cache-miss, and output figures in Chinese-grouped units (e.g. 190万2900, 1亿2900万3999) so there is no mental arithmetic required.

Beyond the numbers, the plugin attaches a mood emoji graded by that round's output tokens, ranging from a relaxed smile up to a full red-face meltdown at the million-token mark, turning raw cost into an instant emotional gauge. Rounds that produced file artifacts keep a clickable output row, and when reasoning tokens are present a hover tooltip expands to show the breakdown of output versus reasoning, so the information stays layered without being noisy.

It is aimed at developers who use the DeepSeek Harness Web profile day-to-day and regularly run long-context or multi-file editing sessions: whether you are managing an API budget, comparing prompt strategies by token efficiency, or simply want a pinch of personality on otherwise bland integers, this small tail line makes the cost of every interaction tangible at a glance.

Use Cases

  • Check per-round input, cache, and output token details at a glance after a multi-turn conversation
  • Compare token efficiency across different prompt strategies or model configurations
  • Monitor single-reply output volume to avoid unexpectedly high API bills

Best For

  • Developers who use DeepSeek Harness Web GUI for multi-turn interactions daily
  • Engineers running long-context or multi-file editing sessions frequently
  • AI application builders who track API costs and token efficiency