AI Agent Hub
Back to plugins
🖥️

dsh-widechat

Client Updated 2026.09.05

Run the following command in DeepSeek Harness:

dsh plugin install FraYoshi/dsh-widechat

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

Install this plugin in DeepSeek Harness by running dsh plugin install FraYoshi/dsh-widechat; the full source is available at https://github.com/FraYoshi/dsh-widechat

About this plugin

The stock DSH Web UI pins chat content to 748px and centers it, so on any monitor above 1024px the reading area is the narrowest element on screen, the user message bubble looks thin on wide viewports, and a long input pushes the conversation off-screen. dsh-widechat is a lightweight client-side patch aimed at exactly these three pain points.

It replaces the fixed pixel cap with a percentage-of-cell layout that reflows automatically as the sidebar collapses to a rail or expands back; lets you switch the session stats strip (token counts, duration, etc.) between left, center, and right alignment; and clamps the composer scroll region between 20 and 80 percent of the viewport so long inputs keep their room without swallowing the conversation above. All four values live on a single row in the General settings tab — three sliders plus a three-way toggle — and update live without a DSH restart, only overriding CSS variable values rather than touching the stock rules.

Built for developers on 2K or 4K displays who find the default column cramped. Be aware the plugin targets six CSS-Module hash class names from dsh-client-ui-conversation; each upstream rebuild can rotate those hashes, in which case the class names need a manual update and a client rebuild per the README.

Use Cases

  • Default DSH chat column feels too narrow with large gutters on a 2K or 4K monitor
  • Long prompts in the composer push the entire conversation off-screen
  • Want to tune bubble width and stats alignment without editing upstream source code

Best For

  • Developers using DSH Web UI daily on wide or ultrawide monitors
  • Users who prefer a custom layout without forking upstream packages
  • AI-assisted development workflows where long prompts are common