Introduction

In the plugin system of DeepSeek Harness (DSH), a common practice is to encapsulate different capabilities into installable, triggerable plugins. For explanatory tasks, plain text output is sometimes not intuitive enough: conceptual relationships, process steps, device structures, etc., are easier to express with diagrams.

dsh-illustrated-teaching is a community plugin for DSH designed to render knowledge points into illustrated galleries and provide local export.

What is this

dsh-illustrated-teaching is a DeepSeek Harness plugin maintained by Henryang777 and released under the MIT license. It automatically renders knowledge points into illustrated galleries, specifically designed for undergraduate-level learning.

Its main approach is: the model plans 3-6 images and fills them with knowledge data; renders them into illustrated cards; generates a self-contained gallery page; and exports to a local folder.

Core Capabilities

  • Provides 8 sets of Xiaohongshu popular science style templates.
  • Supports Harness dialogue triggering and HTTP API invocation.
  • Generates self-contained gallery pages with images inlined as base64.
  • Exports PNG, captions.txt, and manifest.json to a local folder.
  • Supports retrieval of real images from Wikimedia Commons/Wikipedia with automatic attribution.
  • When offline, real images automatically degrade to SVG illustrations without throwing errors or interrupting the process.
  • Rendering fallback chain is: Playwright -> System Chrome/Edge -> Native SVG rendering.

Installation and Enablement

Prerequisites:

  • DeepSeek Harness is installed with version dsh 0.1.0-rc.6.
  • Node version meets node >=22.19.0.

The installation command is as follows:

dsh plugin --profile web add github:Henryang777/dsh-illustrated-teaching

After installation, you need to restart Harness for the plugin to take effect.

If you need to use Playwright rendering, you can first install Chromium:

npx playwright install chromium

You can also use an installed Chrome/Edge, and the plugin will attempt to detect it automatically. The browser path can also be specified via the environment variable DSH_ILLUSTRATED_CHROME_PATH.

The installation verification endpoint is as follows:

curl http://127.0.0.1:3080/illustrated-teaching/health

Typical Usage

In Harness conversations, you can trigger it directly using natural language. Verified examples include:

  • “Explain what a MOSFET is with a diagram”
  • “Explain what a PN junction is with a diagram”
  • “Help me understand the difference between processes and threads in an operating system”
  • “Give a popular science explanation of protein purification processes, with diagrams”
  • “Understand the chip industry chain in one image”

You can also call the HTTP API directly, which is suitable for scripts or automated workflows:

curl -X POST http://127.0.0.1:3080/illustrated-teaching/generate

Applicable Scenarios and Notes

Suitable for scenarios where concepts, processes, structures, and other knowledge points are organized into browsable illustrations. The knowledge data generated by the plugin is provided by the invoked language model; please verify it yourself before using it in serious scenarios.

The visual style is inspired by excellent semiconductor popular science images on Xiaohongshu/Bilibili, borrowing only the layout language without using their image assets.

The plugin runs with the permissions of the current DSH process. You should check the source code and MIT license before installation. The community directory is an independent site with no official affiliation with DeepSeek/Fangfang.

Links

  • GitHub: https://github.com/Henryang777/dsh-illustrated-teaching
  • Directory Page: https://www.skillhub.cn/plugins/Henryang777/dsh-illustrated-teaching