AI Agent Hub
Back to plugins
🖥️

dsh-conversation-folding

Client Updated 2026.08.23

Run the following command in DeepSeek Harness:

dsh plugin install TheHeartFickle/dsh-conversation-folding

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

Run dsh plugin install TheHeartFickle/dsh-conversation-folding in the DeepSeek Harness terminal to install the plugin, whose source repository is https://github.com/TheHeartFickle/dsh-conversation-folding ; Node.js 20 or later and a working dsh web session are required. After installation, restart DSH and refresh the browser page to activate.

About this plugin

In a DSH conversation, thinking traces, tool calls, and final answers all stream into the same scroll, making the feed feel like raw, unedited logs. dsh-conversation-folding tackles exactly that friction: without altering any content, it collapses intermediate steps behind UI panels so that user prompts and LLM output always render in their native style, keeping your eye on what actually matters.

The plugin works on three levels. Round-level folding (the default) gathers a turn's thinking and tool calls into a single collapsible bar; once the main response starts streaming, the bar tucks itself away automatically. Consecutive tool-call steps are further grouped, showing only the latest entry by default to cut down repetition. Chain-of-thought blocks open as scrollable sub-frames that auto-track the bottom during generation, and a new block only folds the immediately preceding one so context stays continuous. Auxiliary lines such as context injection and skill loading can be toggled individually via a simple config key.

If you frequently trigger tool calls or work with long thinking chains in DSH and want your conversation feed to read like a polished article rather than an operation log, this plugin is a good fit. It is non-invasive, ships with few configuration options, and takes effect immediately after a page refresh.

Use Cases

  • Cutting down repeated tool-call noise in a busy conversation feed
  • Tracking long chain-of-thought generation in real time without losing context
  • Reading the conversation like a polished article instead of a raw log

Best For

  • DSH users who fire frequent tool calls in a session
  • Developers who rely on long chain-of-thought for multi-step reasoning
  • Heavy CLI users who value a clean conversation reading experience