Introduction

The philosophy of DSH (DeepSeek Harness) is “everything is a plugin.” The framework itself is only responsible for scheduling, and specific capabilities are supplemented by skill packs. Skills in the community are mostly concentrated in engineering scenarios such as coding and looking up information. If you want your agent to conveniently handle some daily life requests—for example, a user uploads a floor plan asking about Feng Shui, or provides a birthday to calculate their Bazi—it is much easier to install a ready-made skill following the Agent Skills standard than to write calendar conversions and prompts from scratch. Below is an introduction to Big8, maintained by kobenfang.

What is it

Big8 is a skill pack following the Agent Skills open standard (SKILL.md + YAML frontmatter). The author is kobenfang, the license is MIT, the npm package name is @kobenfang/dsh-big8, and the version is 1.0.0. It encapsulates inquiries about Feng Shui, physiognomy, Bazi, zodiac signs, divination, and the Old Almanac into a single skill, suitable for frameworks that support Agent Skills, such as DeepSeek Harness (dsh), OpenClaw, and Claude Code.

Core Features

The verified capabilities are six:

  1. Read image to analyze Feng Shui: Upload an image to analyze home Feng Shui.
  2. Physiognomy analysis: Upload an image to analyze facial features.
  3. Calculate Bazi: Input a birthday to generate the Four Pillars of Destiny (Bazi).
  4. Check zodiac horoscope.
  5. Daily Hexagram / Divination.
  6. Old Almanac.

Complete feature descriptions are in the repository’s SKILL.md. Looking at the directory structure provided in the README, the skill itself consists of a Python script plus four knowledge files:

big8/
├── SKILL.md          # Skill definition (Agent Skills standard)
├── scripts/big8.py
├── scripts/knowledge/bazi.yaml
├── scripts/knowledge/face.yaml
├── scripts/knowledge/fengshui.yaml
├── scripts/knowledge/zodiac.yaml

Installation and Activation

First, install environment dependencies: Python 3 and lunar-python.

pip install lunar-python

Then, clone the repository to a user-level shared directory as recommended by README Method 1:

mkdir -p ~/.agents
git clone https://github.com/kobenfang/big8.git ~/.agents/big8

If you only want to use it in a single project, you can use Method 2 and copy the skill into the project directory:

mkdir -p .agents
cp -r big8 .agents/

After the above steps, big8 can be found in the skill list when dsh starts.

Additionally, OpenClaw users can obtain the same skill from ClawHub:

npx clawhub install big8

The search command is npx clawhub search big8.

Typical Usage

Once the skill is enabled, simply converse in natural language. Examples from the README include:

  1. Upload an image and ask it to analyze home Feng Shui or facial features—this type requires multimodal model support, as text-only models cannot read images.
  2. Input a birthday to calculate Bazi.
  3. Check the zodiac horoscope, or perform divination.

Applicable Scenarios and Notes

It is suitable for two groups of people: developers who want to add fun Q&A capabilities to their dsh agents; and people who want to reference how a standard Agent Skills skill pack is organized (SKILL.md definition + Python script + knowledge YAML). This repository is small in size and has a clear structure, making it suitable as a template.

Note three points before use:

  1. Feng Shui and physiognomy interpretations require uploading images and rely on multimodal models; if the model does not support image reading, these two functions will not work.
  2. The skill runs with the permissions of the current dsh process. It is recommended to review the repository source code before installing—the skill itself consists of SKILL.md, a Python script, and several YAML files, which can be read directly before installation.
  3. Interpretation results are fun content belonging to traditional culture and should not be used as a basis for decision-making.

Conclusion

The value of Big8 lies in encapsulating a type of daily life inquiry into a standard skill pack that can be cloned to a directory for immediate use, and it also serves as a reference for writing your own skills.

  • GitHub Repository: https://github.com/kobenfang/Big8
  • Community Directory Page: https://www.skillhub.cn/plugins/kobenfang/Big8 (This directory is an independent community site and has no official affiliation with DeepSeek / Fangfang)