In development and usage scenarios for DeepSeek Harness (DSH), building clear context is the foundation for ensuring dialogue quality. Developers often face issues such as missing background information, shared blind spots with the AI, or the inability to structure fragmented thoughts into output. To solve these pain points, cyrus123456 developed this plugin, which aims to use the Johari Window model to organize the dialogue context and generate structured prompts.

This is a DSH Web UI plugin that helps users organize the conversation background through a visual four-quadrant panel. The plugin follows the MIT license, and the code is built with TypeScript + React 18.

Core Features

  1. Add button: Adds a “Johari Quadrants” button above the chat input box.
  2. Popup panel: After clicking the button, a 2×2 quadrant panel pops up for users to fill in.
  3. Auto fill: After filling it out and clicking “Generate Prompt and Fill”, the plugin combines the four-quadrant content into a structured prompt and automatically writes it into the chat input box.

Installation and Enablement

It is recommended to install it using DSH’s built-in plugin management tool:

dsh plugin add cyrus123456/dsh-johari-cognition-quadrant-dialog-composer

After installation is complete, restart the DSH process to see the new button in the chat interface.

Usage

The panel contains four quadrants. The horizontal axis indicates “I know / I don’t know”, and the vertical axis indicates “AI knows / AI doesn’t know”:

  • My Hidden Area (top left): I know · AI doesn’t know. Enter background information that needs to be told to the AI here.
  • Shared Unknown Area (top right): I don’t know · AI doesn’t know. Enter topics that need to be explored together with the AI here.
  • Public Consensus Area (bottom left): I know · AI knows. Enter consensus background already shared by both parties here.
  • My Blind Area (bottom right): I don’t know · AI knows. Enter the areas you want the AI to teach or explain here.

After filling in the four quadrants, click the “Generate Prompt and Fill” button on the panel. The system will automatically combine the above content into a structured text, making it convenient to send directly to the AI afterwards.

Technical Details

The plugin is developed based on the DSH Plugin API. Its main tech stack is as follows:

  • Language and framework: TypeScript + React 18
  • Build tool: esbuild (compatible with Node 20+)
  • Style injection: Runtime CSS injection, using the johari- prefix to isolate class names
  • Dependency interface: conversation.input.dock slot

For local development, you need to clone the repository, install dependencies, and build the project first. The build artifacts are located in the lib/ directory, including the Host-side entry index.js and the Client-side entry client.js.

Applicable Scenarios

This tool is suitable for developers and users who need to deeply organize context, perform complex role-playing, or organize prompts in a systematic way. Before use, please confirm that the current DSH version meets the requirements, and it is recommended to check the source code and license before installation.

References

  • Plugin directory: https://www.skillhub.cn/plugins/cyrus123456/dsh-johari-cognition-quadrant-dialog-composer
  • Source repository: https://github.com/cyrus123456/dsh-johari-cognition-quadrant-dialog-composer