Liki: Chinese Metaphysics AI Skill with Verifiable Conclusions

What changed

The Liki project, a professional AI skill for Chinese Metaphysics, has seen a recent release on October 13, 2026. This release marks a fresh update for the Go-based repository, which aims to provide a professional standard for metaphysical practices. The skill covers a range of disciplines including BaZi, ZiWei, Liuyao, QiMen, date selection, Feng Shui, and naming. Key features highlighted include an astronomical-engine charting system, judgments sourced from classical texts, and an emphasis on verifiable conclusions.

Why it matters for builders

For developers interested in AI agents and specialized domains, Liki presents an example of a skill built to professional standards. Its architecture, which relies on an astronomical-engine for charting and provides classical sourcing for its pronouncements, could serve as a reference for creating similar domain-specific AI applications. The project's use of Go and its JSON-RPC capabilities suggest potential integration pathways for other services.

Practical impact

Builders can explore Liki to understand how to structure an AI skill for complex, tradition-bound domains. The project's commitment to verifiable conclusions and classical sourcing offers a model for ensuring accuracy and transparency in AI-driven interpretations. Developers looking to incorporate Chinese Metaphysics into applications, or to build similar specialized AI tools, may find value in examining its implementation and approach. The project has 85 stars and 13 forks on GitHub, indicating some community interest.

Caveats and source limits

Detailed technical documentation beyond the README summary is not available. The repository does not provide explicit examples of usage or clear instructions on setup and requirements, beyond the mention of Go as the primary language. Information regarding specific astronomical libraries used, the exact nature of the "verifiable conclusions" mechanism, or performance benchmarks is absent. The project's description and README are primarily in Chinese, with an English translation provided, which may limit accessibility for some international developers. The provided metadata does not include details on deployment or integration beyond the mention of JSON-RPC.