一个面向多人德州场景的实时策略引擎实验项目。 An experimental real-time multiplayer Texas Hold'em strategy engine.
本项目尝试探索一条不同于传统 CFR 或 solver-first 主链的德州 AI 工程路径。 它不是一个成熟的 GTO 解算器，也不是对商业求解器的再包装，而是一个围绕事件驱动、状态隔离、样本回放、持续验证和 AI 辅助开发逐步构建出来的复杂系统实验样本。
This project explores an alternative poker AI engineering path beyond a traditional CFR-first or solver-first mainline. It is not a mature GTO solver and not a repackaged commercial engine. Instead, it is a complex-system experiment built around event-driven design, state isolation, replay-based validation, continuous debugging, and AI-assisted development.
项目早期轮廓的确立，参考了商业 API 的公开资料与协议材料，尤其包括对 manual_eng(1).md 与 openapi(1).yaml 的分析。
这些材料帮助我理解了外部接口的表达方式，并在中国国内大模型的辅助下确立了系统基础轮廓。
The early outline of this project was informed by the study of commercial API reference materials, especially manual_eng(1).md and openapi(1).yaml.
These materials helped me understand external interface conventions and, together with analysis assisted by domestic Chinese large models, shaped the initial structure of the system.
我始终遵守基本的道德底线，从未对任何商业目标 API 的服务器实施探针活动、侵入式渗透或其他越界行为。 如果在代码构建层面与某些商业系统存在相似之处，那也只是工程实现上的雷同，而非抄袭。
I have always respected basic ethical boundaries. I have never performed probing, intrusive penetration, or any other unauthorized activity against any commercial target API server. If some build-level similarities exist between this project and commercial systems, they should be understood as coincidental convergence in engineering implementation, not plagiarism.
这个项目的价值，不在于宣称“最强策略”，而在于展示：在人类工程约束下，如何借助 AI 工具逐步搭建一个可运行、可回放、可审计、可持续迭代的多人德州策略引擎。 我相信开源不仅是协作方式，也是复杂工程进步的阶梯。即使这个系统今天仍然简陋，它依然代表了一条重要尝试：不完全依赖 CFR 算法，也能在 AI 辅助下推进德州复杂系统的构建。
The value of this repository is not to claim perfect strategy, but to demonstrate how a human, under clear engineering constraints, can use AI tools to build a runnable, replayable, auditable, and iteratively improvable multiplayer poker strategy engine. Even if the system is still rough today, it represents a meaningful attempt: building a poker strategy system without fully depending on CFR solvers, while relying heavily on AI-assisted engineering.
多人德州实时策略引擎 / Real-time multiplayer Texas Hold'em strategy engine