AI Agent Hub
Back to plugins
🖥️

dsh-message-edit

Client Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install TwilightSpirit/dsh-message-edit

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

Run dsh plugin install TwilightSpirit/dsh-message-edit in the DeepSeek Harness terminal, or visit https://github.com/TwilightSpirit/dsh-message-edit to browse the source and manually add the plugin to the bundles list in ~/.dsh/profiles/web/ package.json.

About this plugin

You are mid-conversation, spot a wrong parameter in your prompt or a flawed answer from the model, and want to fix it before continuing. But DSH uses an append-only log, so no old event can be physically removed. dsh-message-edit makes the fix feel natural: click the pencil icon on the bubble, edit the text, hit save, and the model immediately proceeds with the revised context.

Under the hood it does more than swap display text. The plugin issues a surface-replace event through the official DSH API, so every subsequent turn the model sees the corrected version. The original event stays in the append-only log for a complete audit trail. Both user messages and assistant replies are editable; bold, tables, code blocks, and other markdown are rendered in place. Repeated edits on the same message are tracked correctly, and a forked child session inherits all modifications automatically.

Built for engineers and prompt engineers who need precise control over conversation state: correcting a typo in a critical parameter, rewriting a wrong conclusion before asking a follow-up, or iterating a prompt within the same branch. Zero configuration required.

Use Cases

  • Fix a typo in a critical prompt parameter and let the model continue
  • Rewrite a wrong assistant conclusion before asking a follow-up
  • Iteratively refine prompts within the same branch without restarting

Best For

  • Engineers who need fine-grained control over conversation context
  • Prompt engineers iteratively debugging and refining prompts
  • Teams that need to correct dialogue while preserving audit logs