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.
Badge for your README
Markdown
[](/listings/shizhilya--yuan)HTML
<a href="/listings/shizhilya--yuan"><img src="/api/listings/shizhilya--yuan/badge.svg" alt="findsafeskills safety rating" /></a>Install
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)
- ---
- name: yuan
- 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.
- ---
- # Yuan
- 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。
- 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”。
- ## 启动问询
- 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:
- 1. 姓名和性别
- 2. 出生年月日时分
- - 默认要求公历。
- - 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。
- - 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。
- 3. 出生地点
- - 至少到城市。
- - 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。
- 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。
- ## 统一输入
- 信息齐全后,先在内部整理成一个统一输入包:
- ```json
- {
- "name": "",
- "gender": "",
- "birth_datetime": "",
- "calendar_type": "solar|lunar|unknown",
- "birthplace": "",
- "timezone": "",
- "question_focus": []
- }
- ```
- 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。
- 如用户没有额外指定,再补一个内部渲染配置:
- ```json
- {
- "style_mode": "conservative|direct",
- "assertiveness": "low|medium|high",
- "year_window": {
- "anchor_year": 2026,
- "past": 2,
- "future": 2
- }
- }
- ```
- - `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。
- - `style_mode=direct`:直接输出结论、称骨歌诀、分主题断语、五年断事。
- - `assertiveness=high` 仅在用户明确要求“直断、详细、不要摘要”时使用。
- - `year_window` 默认锚定当前年份,输出“前两年 + 当年 + 后两年”共五年。
- ## 六法参考
- 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。
- | 方法 | 参考入口 | 要点 |
- | --- | --- | --- |
- | 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 |
- | 称骨 | `references/chenggu/SKILL.md` | 必须给出称骨重量、歌诀、白话解释 |
- | 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 |
- | 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 |
- | 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 |
- | 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 |
- 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。
- 总 skill 的目标不是机械凑满六法,而是:
- 1. 先运行所有“当前可辩护的方法”
- 2. 明确标注 `runnable / partial / blocked`
- 3. 在用户要求详细直断时,用可运行方法给出成品级输出
- 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
- 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
- ## 执行流程
- 1. 读取必要方法参考。
- 2. 对每个方法生成内部结果卡:
- - `method`
- - `status`
- - `facts`
- - `interpretation`
- - `confidence`
- - `blockers`
- - `evidence`
- 3. 对称骨结果额外保留:
- - `bone_weight`
- - `verse`
- - `plain_language`
- 4. 对可运行方法至少归一到六个维度:
- - 性格 / 底层气质
- - 事业 / 适配场景
- - 财运 / 资源模式
- - 关系 / 亲密模式
- - 身心 / 压力来源
- - 阶段 / 运势节奏
- 5. 如果 `style_mode=direct`,额外生成:
- - `final_verdict`
- - `theme_sections`
- - `five_year_forecast`
- - `actionable_advice`
- 6. 再做一致性判断。
- ## 交付门槛
- 只有满足下面条件,才输出最终成品:
- - 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。
- - 至少三个可运行来源能支撑核心判断。
- - 称骨歌诀已取到。
- 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:
- 1. 出生时间是否准确到分钟,是否可能跨时辰。
- 2. 日期到底是公历还是农历;如农历,是否闰月。
- 3. 出生城市是否准确,是否需要按真太阳时修正。
- 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。
- 每轮最多问 3 个问题。
- 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:
- - 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)
- - 称骨歌诀已取到
- - 无法运行的方法必须写清限制
- - 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`
- - `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
- ## 输出结构
- 无论 `style_mode` 是什么,最终交付都按这个顺序:
- 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。
- 2. `称骨歌诀`:必须完整给出骨重、歌诀、白话直断。
- 3. `结论`:直接写总体命势,不要先写共同结论。
- 4. `事业`
- 5. `财运`
- 6. `感情`
- 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”。
- 8. `建议`:保留现实可执行建议。
- ### 渲染规则
- - 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`。
- - 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。
- - `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。
- - 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。
- - 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)
- # Yuan
- 🌐 **English** · [中文](./README-CN.md)
- `Yuan` is a unified destiny-reading skill for Codex, Claude Code, and any Agent Skills–compatible runtime.
- 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.
- ---
- ## Positioning
- `Yuan` cares about three things:
- - **Unified input** — one set of birth data, no per-system improvisation
- - **Unified delivery** — final results only; no method tour, no divergence analysis, no intermediate scaffolding
- - **Unified boundary** — run what can be run, mark what cannot, never fabricate facts
- In other words, `Yuan` is not a "metaphysics router" — it is a **destiny-reading renderer**.
- ---
- ## Design Principles
- - **Single input surface**
- Birth datetime, birthplace, calendar type, gender, and focus topics are first reduced to one canonical input package, then derived into each reference system.
- - **Final-result delivery**
- 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.
- - **Runnable-first**
- Prefer methods that genuinely hold up in the current environment. Anything `partial` or `blocked` is declared as a limit, not faked.
- - **Zero-dependency by default**
- Input validation, result rendering, and structural constraints work without extra installs. Optional toolchains exist only for precise charting.
- - **Deterministic facts, heuristic reading**
- Calculation and interpretation are strictly layered. Traditional heuristics are allowed, but never packaged as scientific proof.
- ---
- ## Capability Surface
- `Yuan` orchestrates six reference systems under one umbrella skill:
- | Method | Reference |
- | --- | --- |
- | BaZi (Four Pillars) | `references/bazi/` |
- | Cheng Gu (bone-weight) | `references/chenggu/` |
- | Numerology | `references/numerology/` |
- | Western Astrology | `references/western-astrology/` |
- | Vedic Astrology | `references/vedic-astrology/` |
- | Zi Wei Dou Shu | `references/ziwei/` |
- Final delivery focuses on three rendering modes:
- 1. `Conservative` — restrained tone, preserves boundary awareness
- 2. `Direct` — verdicts, judgments, and a five-year window without intermediate analysis
- 3. `Five-year window` — anchored at the current year, two years back and two forward
- ---
- ## Output Contract
- Final delivery is always organized in this order:
- 1. `Input Confirmation`
- 2. `Bone-Weight Verse`
- 3. `Verdict`
- 4. `Career`
- 5. `Wealth`
- 6. `Relationships`
- 7. `Five-Year Forecast`
- 8. `Advice`
- The following sections **never appear** in final output by default:
- - Method overview
- - Cross-method consensus
- - Divergence and confidence
- - Working-process narration
- - "What methods I used" explanations
- ---
- ## Boundaries
- `Yuan`'s boundaries are explicit:
- - Does not present metaphysics as scientific fact
- - Does not fabricate stars, palaces, Nakshatras, or Dashas when reliable charting is unavailable
- - Does not leak intermediate scaffolding as final output
- - Does not require dependency installs for basic interaction
- 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.
- ---
- ## Repository Layout
- ```text
- yuan/
- ├── SKILL.md
- ├── README.md
- ├── README-CN.md
- ├── AGENTS.md
- ├── agents/
- │ └── openai.yaml
- └── references/
- ├── bazi/
- ├── chenggu/
- ├── numerology/
- ├── western-astrology/
- ├── vedic-astrology/
- └── ziwei/
- ```
- ---
- ## Installation
- ### Codex
- Project-level:
- ```bash
- .agents/skills/yuan/
- ```
- User-level:
- ```bash
- $HOME/.agents/skills/yuan/
- ```
- ### Claude Code
- Project-level:
- ```bash
- .claude/skills/yuan/
- ```
- User-level:
- ```bash
- ~/.claude/skills/yuan/
- ```
- ---
- ## Invocation
- Explicit call:
- ```text
- Use the yuan skill to produce a direct final reading with bone-weight verse,
- verdict, career, wealth, relationship, and a five-year window.
- ```
- Natural language call:
- ```text
- Run yuan and give me the final reading directly, no intermediate steps.
- Name: Zhang San
- Gender: Male
- Solar: 1992-08-14 21:30
- Lunar: 1992 / 7th month / 16th, hour of Hai
- Birthplace: Wuhan, Hubei
- ```
- ---
- ## Recommended Evolution
- Three highest-leverage improvements going forward:
- - Add a peer-level `SKILL.md` for BaZi so all six reference entries are isomorphic
- - Document explicit runtime parameters for `style_mode / assertiveness / year_window`
- - Provide clearer fallback templates for `blocked` systems so final delivery stays stable
- 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 inFree to use. If it’s helped, pass it on.