AI Agent Hub
Back to plugins
🖥️

dsh-quiet-mode

Client Updated 2026.09.16

Run the following command in DeepSeek Harness:

dsh plugin install viyiviyi/dsh-quiet-mode

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

First clone the repository from https://github.com/viyiviyi/dsh-quiet-mode to your local machine, then run dsh plugin install viyiviyi/dsh-quiet-mode inside the cloned directory, and restart dsh web to activate.

About this plugin

Long conversations with DeepSeek Harness have a predictable drift: the model starts over-designing, inventing history, coining jargon, and hiding behind acronyms. You end up repeating your instructions every turn, only to watch them get buried a few messages later.

dsh-quiet-mode solves this by injecting a fixed working-discipline directive into the system prompt of every model request under a given profile. Once installed, the directive is always present—no manual repetition, no risk of it getting squeezed out by later turns. Want different wording? Override it at the profile configuration layer; the bundled defaults are never the only source of truth.

It is built for DeepSeek Harness users who want their model to stay focused, concise, and honest—especially workflows that rely on stable Simplified Chinese output and zero tolerance for fabricated context or unnecessary abstraction. Install, restart, and get results instead of rambling.

Use Cases

  • Fixed instructions get buried after a few turns of conversation
  • Manually restating work rules in every new session
  • Need the model to stay in Simplified Chinese and avoid over-engineering

Best For

  • Everyday DeepSeek Harness users in long-form chat
  • Developers who want stable, concise, non-fabricated model output
  • Automation-workflow users who rely on a fixed prompt discipline