@QingQ77: An AI Agent-oriented fate analysis Skill that uses algorithms for accurate chart casting (no LLM guessing), supports three modes: Bazi standalone, Ziwei standalone, and Bazi+Ziwei comprehensive cross-validation, with one-click generation of ink-wash style HTML fate chart posters. https://github.com/dzcmemory-w…

A Fate Analysis Skill for AI Agents: uses algorithms to accurately calculate charts (no LLM guessing). Supports three modes: Bazi standalone, Ziwei standalone, and Bazi + Ziwei cross-verification. One-click generation of ink-wash style HTML chart posters. https://github.com/dzcmemory-web/bazi-ziwei-skill… Can be installed into Claude Code, Codex, Cursor, and other AI Agents. Chart calculation uses the open-source algorithms Yiqi and lunar-typescript, runs locally, no LLM guessing required. Three modes: Bazi standalone, Ziwei standalone, Bazi + Ziwei cross-verification (cross-check between two systems). Output can be a long Markdown article or a single-file HTML ink-wash poster containing the Ziwei 12-palace chart, Bazi Four Pillars chart, and a six-dimension cross-check table. Runs entirely locally, no internet required, privacy-first. — # dzcmemory-web/bazi-ziwei-skill Source: https://github.com/dzcmemory-web/bazi-ziwei-skill # bazi-ziwei-skill AI Bazi + Ziwei Wei Dou Shu Chart Calculation & Cross-Verification Skill Accurate chart calculation (no LLM guessing) · Three analysis modes · One-click generation of ink-wash style HTML chart posters License: MIT SKILL.md 简体中文 | English Example of a cross-verification poster (synthetic life lord, for display only) — ## What is this? A fate analysis Skill following the SKILL.md open standard (https://code.claude.com/docs/en/skills), installable into Claude Code / Claude Desktop / Codex / Cursor / Hermes / OpenClaw and other AI Agents that support this standard. It does three things that large models alone do poorly: 1. Accurate chart calculation: The Bazi Four Pillars, Ziwei 12 Palaces, Great Luck, and Current Year Luck are computed by built-in algorithm libraries. The LLM does not calculate them itself — pure LLM chart calculation often gets the Day Pillar, Day Master, and Structure wrong, and one mistake invalidates the entire reading. 2. Structure enrichment layer: On top of the chart calculation, adds a layer of “Structure / Prosperity-Decline / Climate Adjustment / Punishment-Combination-Harm / Cover-Head-Cut-Foot” algorithms, providing the LLM with evidence-based material for analysis. 3. Cross-verification: Cross-checks the conclusions from the two independent systems of Bazi and Ziwei — whether the main themes align, whether life windows correspond, and which to trust when they conflict. This is the core added value of this Skill over “any LLM + any chart tool.” — ## ✨ Features - 🎯 Algorithmic accuracy: Chart calculation core sourced from the open-source project Yiqi (MIT), verified through real-world testing; enrichment algorithms validated through multi-dimensional regression on 7 cases. - 🧭 Three analysis modes: Bazi standalone / Ziwei standalone / Bazi + Ziwei cross-verification. - 📜 Two presentation formats: Long-form Markdown article + 🎴 Single-file HTML poster (only for cross-verification). - 🎴 Ink-wash chart poster: Modern minimalism × Chinese ink-wash, includes Ziwei 12 palace chart + Bazi Four Pillars chart + six-dimension cross-check table. Can be screenshot and shared. - 🔌 Cross-Agent: One SKILL.md works across multiple mainstream Agents. - 🔒 Privacy-first: All chart calculations done locally, no network required. Output artifacts gitignored by default. — ## 🚀 Installation ### 1. Clone the repository bash git clone https://github.com/dzcmemory-web/bazi-ziwei-skill.git ### 2. Install algorithm layer dependencies bash cd bazi-ziwei-skill/calculator npm install > Requires Node.js >= 18. The only algorithm layer dependency is lunar-typescript (MIT). ### 3. Register with your Agent Place the entire bazi-ziwei-skill/ folder into your Agent’s skills directory: | Agent | skills directory | |—|—| | Claude Code / Claude Desktop | ~/.claude/skills/bazi-ziwei/ | | Codex | ~/.codex/skills/bazi-ziwei/ or referenced inside project AGENTS.md | | Cursor | Referenced inside project .cursor/ rules | | Hermes Agent | ~/.hermes/skills/bazi-ziwei/ | | OpenClaw | Its skills directory / ClawHub local install | The Agent will automatically read SKILL.md and invoke it as needed. — ## 📖 Usage After installation, just tell the Agent your birth time: I was born on January 1, 2000 at 12:00 PM, male. Can you read my chart? The Agent will: 1. Ask which analysis mode you want (Bazi / Ziwei / Cross-verification). 2. For cross-verification, ask whether you want a long article or an HTML poster. 3. Call the algorithm layer to calculate the chart → load the corresponding prompt → output analysis or render the poster. For detailed flow, see SKILL.md. For testing methods, see TEST-GUIDE.md. ### Command-line chart calculation (without Agent) bash cd calculator # Chart calculation → JSON node dist/run-chart.js --year=2000 --month=1 --day=1 --hour=12 --minute=0 --gender=male > chart.json # JSON → readable text chart node dist/dump-text.js --input=chart.json --output=chart.txt # JSON + analysis JSON + template → HTML poster node dist/render.js --chart=chart.json --analysis=analysis.json \ --template=../templates/report-zonghe-poster.html --output=report.html --currentYear=2026 The repository includes a synthetic example (2000-01-01 male, not a real person): - examples/sample-chart.json — algorithm layer chart output - examples/sample-chart.txt — Wenmo Tianji-style text chart - examples/sample-analysis.json — cross-verification analysis (example) - examples/sample-report.html — finished cross-verification poster. Download and open in a browser to preview. — ## 📁 Directory Structure bazi-ziwei-skill/ ├── SKILL.md Skill entry point (Agent reads this) ├── TEST-GUIDE.md Testing guide (5 user paths) ├── calculator/ │ ├── run-chart.ts Chart entry: birth info → JSON │ ├── dump-text.ts JSON → Wenmo Tianji-style text chart │ ├── render.ts JSON + analysis + template → single-file HTML │ ├── yiqi-core/ Core chart algorithm (vendored from Yiqi, MIT) │ └── bazi-enrich/ Enrichment algorithms: Structure/Prosperity-Decline/Climate Adjustment/Punishment-Combination-Harm ├── prompts/ │ ├── bazi-prompt.md Bazi standalone analysis (long article) │ ├── ziwei-prompt.md Ziwei standalone analysis (long article) │ ├── zonghe-yinzheng-prompt.md Cross-verification (long article) │ └── zonghe-poster.md Cross-verification (JSON output for poster) ├── templates/ │ └── report-zonghe-poster.html Cross-verification poster template (placeholders) └── examples/ ├── sample-chart.json Synthetic example chart output └── sample-chart.txt Synthetic example text chart — ## 🏗️ How It Works Birth Info ──> run-chart.ts ──> chart.json ──> dump-text.ts ──> chart.txt (Algorithm layer: accurate chart) (LLM-friendly text) │ ┌───────────────────────────┤ ▼ ▼ Long-form prompt Poster prompt (Markdown prose) (Strict JSON output) │ render.ts + template ▼ Single-file HTML poster Key Design: The LLM is only responsible for “analysis,” not for “chart calculation” or “drawing the HTML.” Chart calculation is handled by deterministic algorithms, HTML visuals by fixed templates, and the LLM’s structured output fills template slots—each does its own job without polluting the others. — ## 🙏 Acknowledgements - Core chart algorithm derived from Yiqi Bazi Ziwei chart system (https://github.com/fdxuyq/Yiqi-BaZi-ZiWei) (MIT license), see NOTICE. - Lunar calendar conversion depends on lunar-typescript (https://github.com/6tail/lunar-typescript) (MIT). — ## 📬 Contact For issues, collaboration, or discussion, email: [email protected] If this project helps you, feel free to give it a ⭐ Star. — ## ⚠️ Disclaimer This project is based on the theoretical frameworks of traditional Bazi and Ziwei Wei Dou Shu. It is intended for cultural research and entertainment purposes only. It does not constitute any decision basis for medical, investment, marriage, legal, or other matters. Fate is shaped by personal choices and objective circumstances. — ## 📄 License MIT © 2026 dzcmemory-web