AI Agent Hub
Back to plugins
🖥️

dsh-peak-badge

Client Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install mj5809/dsh-peak-badge

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

Run dsh plugin install mj5809/dsh-peak-badge in your terminal; the source is available at https://github.com/mj5809/dsh-peak-badge . After installation, restart dsh web and hard-refresh the browser.

About this plugin

Since August 2026, DeepSeek has applied peak/off-peak pricing: token rates double during the 09:00-12:00 and 14:00-18:00 Beijing-time windows and are halved outside them. For developers who drive heavy prompt volume through the DSH web GUI every day, knowing which window you are in at a glance directly affects per-conversation cost.

dsh-peak-badge drops a draggable floating pill into the DSH browser. It refreshes every 30 seconds via the Asia/Shanghai timezone, showing a green label for off-peak (half price) or an orange one for peak (2x price). Press and drag to reposition; the location persists in localStorage across page reloads. A single click opens a detail card with the current Beijing time, the two peak windows, and reference input/output prices for DeepSeek V4-Flash and V4-Pro. The plugin is purely client-side with a no-op host half, so it introduces no host-level risk and requires zero configuration.

It is built for developers and power users who chat with DeepSeek through the DSH browser and want to confirm the current rate at a glance before committing to a long generation task, choosing instead to defer expensive work to the cheaper off-peak hours.

Use Cases

  • Check peak vs off-peak status at a glance before starting a long conversation in the DSH browser
  • Schedule expensive generation tasks to the cheaper off-peak window to reduce token spend
  • Complement the AGENTS.md text-rule reminder in basecamp-style repositories

Best For

  • Developers who drive heavy DeepSeek prompt volume through the DSH web GUI daily
  • Individual users who want to time expensive generation tasks to the off-peak window
  • Team engineers managing multi-model workloads in basecamp-style repositories