Skip to content

YYTbit/dsh-plugin-cost-tracker

30Last commit Aug 13, 2026

dsh-plugin-cost-tracker DSH plugin

This plugin tracks token usage per session and calculates costs based on model pricing. It can inject cost awareness into the system prompt to help manage spending. A CLI tool is also provided for quick cost estimation.

How to install the dsh-plugin-cost-tracker DSH plugin

dsh plugin --profile your-profile add dsh-plugin-cost-tracker

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

dsh-plugin-cost-tracker DSH plugin data source

dsh-plugin-cost-tracker DSH plugin snapshot date: Aug 16, 2026

discovered

What the dsh-plugin-cost-tracker DSH plugin can do

  • Tracks token usage per session
  • Calculates costs based on model pricing
  • Injects cost awareness into the system prompt
  • Provides CLI tool for cost estimation

Where the dsh-plugin-cost-tracker DSH plugin fits

  • Monitor token costs during development with DeepSeek Harness
  • Set budget warnings to avoid overspending
  • Estimate costs before running a model using the CLI
  • Integrate cost tracking into automated workflows

Who the dsh-plugin-cost-tracker DSH plugin is for

  • Developers using DeepSeek Harness who want to manage token costs
  • Teams that need to track and optimize AI spending

dsh-plugin-cost-tracker DSH plugin limitations

  • Only supports the listed models (deepseek, gpt, claude, gemini variants)
  • Requires manual configuration of default model and budget warning settings
  • Plugin is relatively new and may have limited compatibility with older DSH versions

dsh-plugin-cost-tracker DSH plugin: from the repository README

Quoted from the YYTbit/dsh-plugin-cost-tracker README, the upstream source of the dsh-plugin-cost-tracker DSH plugin. Copyright remains with the original authors.

Token cost tracker for DeepSeek Harness. ## What it does - Tracks token usage per session - Calculates costs based on model pricing - Injects cost awareness into system prompt - Provides CLI tool for cost estimation ## Install ```sh dsh plugin --profile your-profile add dsh-plugin-cost-tracker ``` ## CLI ```sh cost --model deepseek-v4 --input 10000 --output 5000 # Model: deepseek-v4 # Input: 10,000 tokens @ $0.27/M # Output: 5,000 tokens @ $1.10/M # Cost: $0.01 ``` ## Supported models - deepseek-v4, deepseek-v4-flash, deepseek-r1, deepseek-vl2 - gpt-4o, gpt-4o-mini, gpt-5.2 - claude-sonnet-4, claude-opus-4, claude-haiku-4 - gemini-2.5-pro, gemini-2.5-flash ## Configuration ```yaml - id: cost-tracker name: dsh-plugin-cost-tracker config: defaultModel: deepseek-v4 budgetWarning: 5.0 enableInjection: true ``` ## License MIT -- YYTbit

Read the full READMERepository license: MIT

dsh-plugin-cost-tracker DSH plugin questions

How do I install dsh-plugin-cost-tracker?

Run `dsh plugin --profile your-profile add dsh-plugin-cost-tracker` in your terminal. Make sure you have DeepSeek Harness installed and a profile configured.

How do I configure the default model and budget warning?

Add the plugin configuration to your DSH profile YAML file as shown in the README. Set `defaultModel` to your preferred model and `budgetWarning` to a dollar amount (e.g., 5.0). The plugin will warn when costs exceed that threshold.

Can I add custom models not listed in the README?

The plugin currently only supports the models listed in the README. You cannot add custom models via configuration. If you need a model not listed, check for updates or consider contributing to the plugin's repository.

Does the CLI tool work without configuring the plugin first?

The CLI tool can be used standalone with the `cost` command, but you need to specify the model and token counts. It does not require prior plugin configuration, but it won't track session costs automatically without the plugin being active.

How does the cost injection into system prompt work?

When `enableInjection` is set to true in the configuration, the plugin appends a cost-awareness message to the system prompt each session. This helps the model understand budget constraints, but it may affect response behavior.