AI Agent Hub
Back to plugins
🖥️

dsh-conversation-cost

Client Updated 2026.08.28

Run the following command in DeepSeek Harness:

dsh plugin install WinnieJQ/dsh-conversation-cost

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

Run dsh plugin install WinnieJQ/dsh-conversation-cost in your DeepSeek Harness terminal (source: https://github.com/WinnieJQ/dsh-conversation-cost), then restart dsh web to activate the badge.

About this plugin

Running long multi-turn conversations in DeepSeek Harness raises a deceptively simple question: how much did this session actually cost? The usual answer was to pause your work, open the DeepSeek billing page, and dig through line-item charges. dsh-conversation-cost removes that friction entirely—a cost badge now sits right beside the conversation title in the web UI, and hovering over it unfolds a step-by-step breakdown by model, token type, and peak/off-peak window, all without leaving the page.

Under the hood the plugin keeps things precise yet lightweight. Every LLM step is folded into a per-step usage record (input, cache-hit, and output tokens), linked to the model that actually executed it, and priced against the official rate card split by peak and off-peak windows. The badge and the hover card both show the amount in CNY and USD, with the exchange rate and per-model prices fully configurable. There are zero runtime dependencies and no build step: the host side is pure ESM and the browser side uses a hand-written module factory, so it drops in and starts updating the badge the moment each step completes.

If you run heavy daily workloads through Harness, want to keep a running eye on your API spend, or share a team quota and need transparent per-session accounting, this lightweight badge is the simplest cost dashboard you can add to your workflow.

Use Cases

  • Check cumulative session spend at a glance during long multi-turn web conversations
  • Quickly identify per-session cost when a team shares an API quota
  • Compare actual spend across models and peak/off-peak windows to inform model choice

Best For

  • Developers who spend most of their day in the DeepSeek Harness web UI
  • Engineering teams that need per-session cost visibility to stay within API budgets
  • AI application developers who swap models and context lengths and care about token-level spend