AI Agent Hub
Back to plugins
🖥️

dsh-sidechat

Client Updated 2026.08.22

Run the following command in DeepSeek Harness:

dsh plugin install xiaozhiaixue/dsh-sidechat

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

Run dsh plugin install xiaozhiaixue/dsh-sidechat in DeepSeek Harness to install the plugin; source code is at https://github.com/xiaozhiaixue/dsh-sidechat

About this plugin

Long agent conversations in DSH often demand follow-up questions, option comparisons, or independent reasoning on a specific passage—all without derailing the main agent's current task. dsh-sidechat opens a dedicated agent session in a right-side panel: scoped to the conversation that spawned it, archived immediately, and invisible in the session list—effectively a sidecar thinking cockpit for the main flow.

Every side chat inherits the launching conversation's toolset, model, thinking effort, permission preset, and working directory, and each can be tuned independently. Select text and choose Ask in side chat to stage the selection in the input (nothing is sent until you press Enter); hit the + button for a fresh empty chat in one click. When you are done, insert a whole reply or a highlighted fragment directly into the main composer draft, or summarize-and-insert it as a collapsed context row. If the main agent pops a question dialog, you can bring the question and options into the side chat for analysis before delivering the answer back to the main flow.

The panel is resizable from 280 to 720 px, collapsible, and follows the DSH language setting for zh/en UI. If you use agents for deep research, option comparison, or multi-step reasoning, dsh-sidechat keeps the main thread and the thinking thread running in parallel without blocking each other.

Use Cases

  • Dig into a specific output with follow-up questions without derailing the main agent
  • Explore option A / B / C in parallel side chats, then bring the best line back
  • Analyze a question dialog and its options in a side chat before committing an answer

Best For

  • Researchers and developers running long multi-turn DSH conversations
  • Powerful agent users who need a main task and a side line of thought in parallel
  • Teams that treat agents as collaborators and keep the main thread unblocked