AI Agent Hub
Back to plugins
⚙️

harness-plugin

Workflow Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install Smith-yue/harness-plugin

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

Install harness-plugin by running dsh plugin install Smith-yue/harness-plugin, or clone https://github.com/Smith-yue/harness-plugin and run npm run install:desktop, then restart Desktop.

About this plugin

DeepSeek Harness Desktop is solid functionally but visually plain, and there is no built-in at-a-glance token-cost or cache-hit view for the current session. harness-plugin layers a theming system and a live usage panel on top without touching the native stats bar or duplicating message content.

Two bundled skins (Deep-Night dark, Warm-Paper light) switch instantly in Settings. Five curated photographic backgrounds ship in the package, or you can point the plugin at any local folder for an image carousel with recursive sub-folder traversal (up to 2000 images), a configurable 1-1440 minute interval, pause and next controls. Image opacity and a colour/gradient overlay opacity are independently adjustable; every preference persists to the local settings file and is restored on restart. The Token Usage panel reads the official tokenUsage session projection and shows billed input, output, cache reads, cache writes, and hit rate in real time with no extra database.

Ideal for Harness power users who care about desk aesthetics and token economics. The plugin makes no external network requests, never stores conversation content, reads images only from a user-authorised local directory with path-traversal and symlink-escape rejection, and is released under the MIT licence.

Use Cases

  • Switch to a dark theme during long sessions to reduce eye strain
  • Rotate personal photography as a local desktop background for a personalized workspace
  • Monitor real-time token spend and cache-hit ratio to keep costs in check

Best For

  • Developers who rely on DeepSeek Harness Desktop daily
  • AI users who track token cost and cache efficiency
  • Creators who value a personalized desktop visual environment