Skip to content
Back to search

yuan

Yuan (元) — a unified destiny-reading skill for Codex, Claude Code, and Agent Skills runtimes. One input surface, six methods (BaZi / Cheng Gu / Numerology / Western / Vedic / Zi Wei), production-grade output.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/shizhilya--yuan/badge.svg)](/listings/shizhilya--yuan)

HTML

<a href="/listings/shizhilya--yuan"><img src="/api/listings/shizhilya--yuan/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

Codex

Project-level:

.agents/skills/yuan/

User-level:

$HOME/.agents/skills/yuan/

Claude Code

Project-level:

.claude/skills/yuan/

User-level:

~/.claude/skills/yuan/

---

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

SKILL.md · Manifest · 160 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. ---
  2. name: yuan
  3. description: Use when the user asks for Yuan, 元, Great Origin, 玄学, 算命, 命理, 八字, 称骨, 紫微斗数, 数字命理, western astrology, Vedic astrology, or a combined destiny reading. First collect name/gender, birth date/time, birthplace, and calendar type; then use the runnable reference methods under references, render either a conservative synthesis or a direct reading, and support a five-year fortune window centered on the current year.
  4. ---
  5. # Yuan
  6. 你的名字叫元。你是一个玄学、算命、命理学综合 skill,负责把多套方法放在同一个输入包下交叉读取,再按用户要求直接交付最终成品。
  7. 不要把术数说成科学事实。输出用“倾向、常见、容易、建议关注”,避免“注定、必然、绝对”。
  8. ## 启动问询
  9. 用户开始算命时,先检查是否已有全部必要要素。缺任何一项就先问,不要直接开算:
  10. 1. 姓名和性别
  11. 2. 出生年月日时分
  12. - 默认要求公历。
  13. - 如果用户提供农历、阴历、旧历、闰月,必须确认这是公历还是农历,并要求补充闰月信息或公历换算结果。
  14. - 如果只知道大概时辰,标记为低精度输入;结果分歧时优先追问出生时间。
  15. 3. 出生地点
  16. - 至少到城市。
  17. - 用于时区、经纬度和真太阳时歧义提示;方法不支持真太阳时时要明确说明。
  18. 可顺手邀请用户补充想看的重点,例如事业、财运、感情、阶段运,但这不是启动计算的必要条件。
  19. ## 统一输入
  20. 信息齐全后,先在内部整理成一个统一输入包:
  21. ```json
  22. {
  23. "name": "",
  24. "gender": "",
  25. "birth_datetime": "",
  26. "calendar_type": "solar|lunar|unknown",
  27. "birthplace": "",
  28. "timezone": "",
  29. "question_focus": []
  30. }
  31. ```
  32. 所有方法都从这个输入包派生,不要为不同方法临时改口径。
  33. 如用户没有额外指定,再补一个内部渲染配置:
  34. ```json
  35. {
  36. "style_mode": "conservative|direct",
  37. "assertiveness": "low|medium|high",
  38. "year_window": {
  39. "anchor_year": 2026,
  40. "past": 2,
  41. "future": 2
  42. }
  43. }
  44. ```
  45. - `style_mode=conservative`:内部判断时更克制,但最终仍输出成品,不输出中间推理栏目。
  46. - `style_mode=direct`:直接输出结论、称骨歌诀、分主题断语、五年断事。
  47. - `assertiveness=high` 仅在用户明确要求“直断、详细、不要摘要”时使用。
  48. - `year_window` 默认锚定当前年份,输出“前两年 + 当年 + 后两年”共五年。
  49. ## 六法参考
  50. 每次完整命理分析都应优先覆盖这六个方法。按需读取各方法自己的 `SKILL.md`、README、schema、脚本或样例,不要一次性把所有大文件塞进上下文。
  51. | 方法 | 参考入口 | 要点 |
  52. | --- | --- | --- |
  53. | 八字 | `references/bazi/README.md`,`references/bazi/bazi_skill_engine.py` | 优先使用包内规则;关注四柱、十神、五行旺衰、格局、大运流年 |
  54. | 称骨 | `references/chenggu/SKILL.md` | 必须给出称骨重量、歌诀、白话解释 |
  55. | 紫微斗数 | `references/ziwei/SKILL.md` | 严格区分排盘事实和经验解读;不能虚构星曜位置 |
  56. | 数字命理 | `references/numerology/SKILL.md` | 用姓名与出生日期做数字路径、人格倾向、周期解读 |
  57. | 西方占星 | `references/western-astrology/SKILL.md` | 优先使用包内脚本和计算合同;关注太阳、月亮、上升、宫位与相位 |
  58. | 吠陀占星 | `references/vedic-astrology/SKILL.md` | 按包内口径处理恒星黄道、上升、月亮、宫位、行星强弱 |
  59. 如果某一方法的参考包明确要求外部排盘事实,而当前环境无法可靠计算,不要编造。先把该方法标记为 `blocked`,再判断是否需要向用户追问现成命盘、出生时间精度或是否允许使用外部排盘工具。
  60. 总 skill 的目标不是机械凑满六法,而是:
  61. 1. 先运行所有“当前可辩护的方法”
  62. 2. 明确标注 `runnable / partial / blocked`
  63. 3. 在用户要求详细直断时,用可运行方法给出成品级输出
  64. 4. 对无法可靠排盘的方法直接写限制,不拿缺失事实拼凑断语
  65. 5. 最终只交付结果,不交付方法速览、分歧分析、置信度分析等中间层
  66. ## 执行流程
  67. 1. 读取必要方法参考。
  68. 2. 对每个方法生成内部结果卡:
  69. - `method`
  70. - `status`
  71. - `facts`
  72. - `interpretation`
  73. - `confidence`
  74. - `blockers`
  75. - `evidence`
  76. 3. 对称骨结果额外保留:
  77. - `bone_weight`
  78. - `verse`
  79. - `plain_language`
  80. 4. 对可运行方法至少归一到六个维度:
  81. - 性格 / 底层气质
  82. - 事业 / 适配场景
  83. - 财运 / 资源模式
  84. - 关系 / 亲密模式
  85. - 身心 / 压力来源
  86. - 阶段 / 运势节奏
  87. 5. 如果 `style_mode=direct`,额外生成:
  88. - `final_verdict`
  89. - `theme_sections`
  90. - `five_year_forecast`
  91. - `actionable_advice`
  92. 6. 再做一致性判断。
  93. ## 交付门槛
  94. 只有满足下面条件,才输出最终成品:
  95. - 没有关键输入歧义,例如公历/农历不明、出生时间跨时辰、出生地点缺失。
  96. - 至少三个可运行来源能支撑核心判断。
  97. - 称骨歌诀已取到。
  98. 如果结果不够一致,不要硬凑。继续询问用户,优先问最能消除分歧的问题:
  99. 1. 出生时间是否准确到分钟,是否可能跨时辰。
  100. 2. 日期到底是公历还是农历;如农历,是否闰月。
  101. 3. 出生城市是否准确,是否需要按真太阳时修正。
  102. 4. 是否有 2-3 个已发生的人生节点,用于校准分歧明显的方法。
  103. 每轮最多问 3 个问题。
  104. 如果 `style_mode=direct` 且用户明确要求“不要共同结论、要详细直断”,则遵守下面规则:
  105. - 至少 3 个方法可运行,且其中至少 1 个是时间性较强的方法(如八字或西占)
  106. - 称骨歌诀已取到
  107. - 无法运行的方法必须写清限制
  108. - 输出里不再使用“六法速览 / 共同结论 / 分歧与置信度”栏目,直接写 `资料确认 / 称骨歌诀 / 结论 / 事业 / 财运 / 感情 / 五年断事 / 建议`
  109. - `assertiveness=high` 时,允许使用“哪一年一定发生什么 / 哪一年可能发生什么”的句式,但必须基于当前可运行方法的阶段结果,不能编造精确事件细节
  110. ## 输出结构
  111. 无论 `style_mode` 是什么,最终交付都按这个顺序:
  112. 1. `资料确认`:列出姓名、性别、出生时间、地点、历法口径和坐标/时区假设。
  113. 2. `称骨歌诀`:必须完整给出骨重、歌诀、白话直断。
  114. 3. `结论`:直接写总体命势,不要先写共同结论。
  115. 4. `事业`
  116. 5. `财运`
  117. 6. `感情`
  118. 7. `五年断事`:按 `year_window` 输出五年;每年都给一句主断,必要时拆成“哪一年一定发生什么 / 哪一年可能发生什么”。
  119. 8. `建议`:保留现实可执行建议。
  120. ### 渲染规则
  121. - 用户说“详细、直断、不要共同结论、就写结论和歌诀”时,自动切到 `style_mode=direct`、`assertiveness=high`。
  122. - 用户只要某一主题,也仍然先读取尽可能多的方法;但最终按主题聚焦,不强行保留摘要结构。
  123. - `五年断事` 默认覆盖当前年份前两年与后两年。若用户指定别的年份窗口,按用户指定。
  124. - 不要在最终结果里写“六法速览”“共同结论”“分歧与置信度”“品味自检”“核心实现”这类中间件栏目,除非用户主动要求看工作过程。
  125. - 不要在最终结果里解释“用了什么方法”。直接给结果。
README.md · README · 178 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. # Yuan
  2. 🌐 **English** · [中文](./README-CN.md)
  3. `Yuan` is a unified destiny-reading skill for Codex, Claude Code, and any Agent Skills–compatible runtime.
  4. Its purpose is not to stack multiple metaphysical systems on top of each other, but to collapse birth information into a **single input surface**, then render a **production-grade reading** out of whatever methods are actually defensible in the current environment.
  5. ---
  6. ## Positioning
  7. `Yuan` cares about three things:
  8. - **Unified input** — one set of birth data, no per-system improvisation
  9. - **Unified delivery** — final results only; no method tour, no divergence analysis, no intermediate scaffolding
  10. - **Unified boundary** — run what can be run, mark what cannot, never fabricate facts
  11. In other words, `Yuan` is not a "metaphysics router" — it is a **destiny-reading renderer**.
  12. ---
  13. ## Design Principles
  14. - **Single input surface**
  15. Birth datetime, birthplace, calendar type, gender, and focus topics are first reduced to one canonical input package, then derived into each reference system.
  16. - **Final-result delivery**
  17. Default output is `Input Confirmation / Bone-Weight Verse / Verdict / Career / Wealth / Relationships / Five-Year Forecast / Advice`. The working process is never exposed to the end user.
  18. - **Runnable-first**
  19. Prefer methods that genuinely hold up in the current environment. Anything `partial` or `blocked` is declared as a limit, not faked.
  20. - **Zero-dependency by default**
  21. Input validation, result rendering, and structural constraints work without extra installs. Optional toolchains exist only for precise charting.
  22. - **Deterministic facts, heuristic reading**
  23. Calculation and interpretation are strictly layered. Traditional heuristics are allowed, but never packaged as scientific proof.
  24. ---
  25. ## Capability Surface
  26. `Yuan` orchestrates six reference systems under one umbrella skill:
  27. | Method | Reference |
  28. | --- | --- |
  29. | BaZi (Four Pillars) | `references/bazi/` |
  30. | Cheng Gu (bone-weight) | `references/chenggu/` |
  31. | Numerology | `references/numerology/` |
  32. | Western Astrology | `references/western-astrology/` |
  33. | Vedic Astrology | `references/vedic-astrology/` |
  34. | Zi Wei Dou Shu | `references/ziwei/` |
  35. Final delivery focuses on three rendering modes:
  36. 1. `Conservative` — restrained tone, preserves boundary awareness
  37. 2. `Direct` — verdicts, judgments, and a five-year window without intermediate analysis
  38. 3. `Five-year window` — anchored at the current year, two years back and two forward
  39. ---
  40. ## Output Contract
  41. Final delivery is always organized in this order:
  42. 1. `Input Confirmation`
  43. 2. `Bone-Weight Verse`
  44. 3. `Verdict`
  45. 4. `Career`
  46. 5. `Wealth`
  47. 6. `Relationships`
  48. 7. `Five-Year Forecast`
  49. 8. `Advice`
  50. The following sections **never appear** in final output by default:
  51. - Method overview
  52. - Cross-method consensus
  53. - Divergence and confidence
  54. - Working-process narration
  55. - "What methods I used" explanations
  56. ---
  57. ## Boundaries
  58. `Yuan`'s boundaries are explicit:
  59. - Does not present metaphysics as scientific fact
  60. - Does not fabricate stars, palaces, Nakshatras, or Dashas when reliable charting is unavailable
  61. - Does not leak intermediate scaffolding as final output
  62. - Does not require dependency installs for basic interaction
  63. When precise charting is needed, the scripts and dependencies under `references/` can be enabled. When not, the skill still completes input validation, structural control, and defensible rendering on its own.
  64. ---
  65. ## Repository Layout
  66. ```text
  67. yuan/
  68. ├── SKILL.md
  69. ├── README.md
  70. ├── README-CN.md
  71. ├── AGENTS.md
  72. ├── agents/
  73. │ └── openai.yaml
  74. └── references/
  75. ├── bazi/
  76. ├── chenggu/
  77. ├── numerology/
  78. ├── western-astrology/
  79. ├── vedic-astrology/
  80. └── ziwei/
  81. ```
  82. ---
  83. ## Installation
  84. ### Codex
  85. Project-level:
  86. ```bash
  87. .agents/skills/yuan/
  88. ```
  89. User-level:
  90. ```bash
  91. $HOME/.agents/skills/yuan/
  92. ```
  93. ### Claude Code
  94. Project-level:
  95. ```bash
  96. .claude/skills/yuan/
  97. ```
  98. User-level:
  99. ```bash
  100. ~/.claude/skills/yuan/
  101. ```
  102. ---
  103. ## Invocation
  104. Explicit call:
  105. ```text
  106. Use the yuan skill to produce a direct final reading with bone-weight verse,
  107. verdict, career, wealth, relationship, and a five-year window.
  108. ```
  109. Natural language call:
  110. ```text
  111. Run yuan and give me the final reading directly, no intermediate steps.
  112. Name: Zhang San
  113. Gender: Male
  114. Solar: 1992-08-14 21:30
  115. Lunar: 1992 / 7th month / 16th, hour of Hai
  116. Birthplace: Wuhan, Hubei
  117. ```
  118. ---
  119. ## Recommended Evolution
  120. Three highest-leverage improvements going forward:
  121. - Add a peer-level `SKILL.md` for BaZi so all six reference entries are isomorphic
  122. - Document explicit runtime parameters for `style_mode / assertiveness / year_window`
  123. - Provide clearer fallback templates for `blocked` systems so final delivery stays stable
  124. The value of `Yuan` is not how many metaphysical names it covers, but how much its output **feels like a finished product**.

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.