Zengshan Buei Six-Line Divination Agent

Problem It Solves

Traditional Six-Line divination relies on many mnemonics and domain heuristics. If a model answers freely, it can easily mix schools, skip steps, or fabricate hexagram data. liuyao turns the Zengshan Buei framework into an inspectable workflow: define the question, select the useful spirit, judge strength, map the four generative/controlling forces, check special patterns, determine timing, and produce a structured report. It also stores each reading as case JSON for calibration and review.

How It Works

• Layered questioning: asks 1-2 context-dependent questions by category, such as investment versus salary, or acute illness versus chronic illness.
• Casting and assembly: after receiving six numbers, calls scripts/zhuanghua.py to generate the hexagram instead of calculating it ad hoc.
• Calibration validation: uses the first four steps to infer past conditions and asks the user to confirm options, with up to three retry angles.
• Rule retrieval: consults matching files under knowledge/rules/ for useful spirit, strength, chong/he, hidden spirit, and timing; falls back to scripts/faiss_search.py when needed.
• Case memory: writes casting, calibration, analysis, and feedback into case JSON files to support Cold/Warm/Hot growth and reviews.

Boundaries and Caveats

The skill is limited to Six-Line divination and knowledge management; it should not be treated as medical diagnosis or legal advice. When the data is unclear, it should say so rather than force a reading. Outputs should preserve confidence and avoid promising certainty.