Your LLM Skill can't do astronomy: why packaged divination Skills compute the wrong answer
There's been a wave of Chinese-divination Skills on GitHub this year. The biggest BaZi (Four Pillars astrology) one has ~2.5k stars — interactive chart casting against nine classical texts, plus MBTI mapping, compatibility, and elemental remedies. Another has ~800 stars and leads with engineered hallucination prevention: fixed casting steps, a structured knowledge base, external scripts. Below…
Recent years have seen a surge of Chinese divination skills on GitHub, with one BaZi (Four Pillars astrology) project boasting over 2,500 stars. These skills provide interactive chart casting, MBTI mapping, compatibility analysis, and elemental remedies. However, a closer look reveals that the core arithmetic, which determines the accuracy of the results, is often flawed.
One significant issue is the handling of the year boundary, which is based on solar terms rather than a fixed calendar date. The year changes when the sun reaches 315° of ecliptic longitude, an event that occurs at a specific moment in time. This means that individuals born just before and after this moment can have different year and month pillars, leading to incorrect interpretations.
Another problem is the reliance on a fixed clock time rather than the local solar time based on the birthplace. For example, two cities in China, with different time zones and longitude, can have different hour pillars for the same clock time due to solar corrections and the equation of time. This discrepancy arises because the BaZi system uses the local solar time of the birthplace, not a standardized time zone.
Additionally, the hour that never existed during daylight saving time transitions poses another challenge. For instance, in Shanghai during the 1980s, daylight saving time shifted an hour forward, creating a non-existent hour that could affect the hour pillar calculation. These errors do not trigger errors or warnings, as the system quietly produces the eight-character chart.
Consequently, even when fed to a skilled LLM, the flawed chart will be presented as plausible and well-reasoned, without any indication of its inaccuracies.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.