AI Agent Hub
Back to plugins
🖥️

dsh-interrupt-send

Client Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install Hyna-hla/dsh-interrupt-send

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

Run dsh plugin install Hyna-hla/dsh-interrupt-send in the DeepSeek Harness terminal to install; the full source code is available at https://github.com/Hyna-hla/dsh-interrupt-send

About this plugin

There is a moment in every AI conversation where the model is still grinding through a long response and you have already typed your next question, yet all you can do is wait for it to finish and queue your message. dsh-interrupt-send turns that pause into a single button. It sits in the input toolbar, just to the left of the send key, and only materialises when the session is running and the draft is non-empty, so it never clutters the UI or gets in the way of ordinary typing.

One click does two things in sequence: it calls session.cancel() to abort the current turn, letting the Host settle into a quiet state while preserving anything already queued, and then immediately submits your draft through the standard input pipeline so it is sent, cleared, and recorded in the undo history. If the cancellation step fails for any reason, the plugin gracefully falls back to a normal queued submit, guaranteeing that nothing you typed is ever lost. The regular Enter-key behaviour (queue or steer) is completely untouched; the button is always just an optional fast lane.

It is built for users who move fast in multi-turn conversations—developers iterating on code, researchers chasing follow-ups, or anyone who habitually types the next prompt while the AI is still working and wants to cut in without waiting.

Use Cases

  • The AI is still grinding through a long response and your next question is already typed
  • Rapid iterative debugging where you need to cut in and redirect frequently
  • Your draft is ready but you do not want to wait for the current answer to finish

Best For

  • Developers who habitually draft the next prompt while the AI is still working
  • Users in multi-turn code generation or long-form scenarios who interrupt and follow up often
  • Efficiency-focused conversationalists who dislike waiting in the input queue