In the DeepSeek Harness (DSH) ecosystem, plugins are the core of extending capabilities. For the time-consuming “writing reports” and “creating presentations” tasks in asset pricing and machine learning research, z-asset/za_ppt provides a solution for generating presentation slides from papers.

What It Is

This is a DSH plugin for empirical asset pricing and machine learning/deep learning research. It focuses on converting research paper content into structured presentation materials, supporting the generation of slides in Beamer or Quarto format.

Core Features

The plugin includes the following five core capabilities:
* Literature review
* Data processing
* Model training
* Report writing
* Presentation creation

Installation

There are two installation methods; choose one based on your environment:

dsh plugin --profile <你的profile> add zresearch-ppt

Or install with npm:

npm install zresearch-ppt

Typical Usage

After installation, trigger the plugin as needed in a DSH session. It automatically loads the corresponding skill based on the task. Common trigger phrases include:

  • “Help me conduct a literature review / find relevant papers”
  • “Help me find data / evaluate datasets”
  • “Run regressions / train ML models / predict cross-sectional returns”
  • “Write a paper / draft sections”
  • “Create presentation slides / Beamer slides / a job market talk”

Technical Details

The plugin contains one skill internally, with the following directory structure:

skills/zresearch-ppt/
├── SKILL.md              # 主 skill 定义
└── references/           # 角色定义 + 领域校准
    ├── agents/           # worker/critic 角色
    └── domain/           # domain-profile(资产定价+ML/DL)

Its domain calibration focuses on empirical asset pricing and machine learning/deep learning, covering scenarios such as factor models, cross-sectional return prediction, GMM/SDF, and predictive regressions.

Dependencies and License

  • Dependencies: @deepseek-ai/cordis and @deepseek-ai/dsh-skill
  • License: MIT

Use Cases and Notes

It is suitable for users who need to conduct empirical asset pricing or machine learning/deep learning research. Before use, check the source code and license (MIT), and make sure it runs with the current DSH process permissions.

As part of the Z-Asset research workflow, this plugin aims to automate the document-generation part of the research process. For the project repository and more details, see GitHub.