中国法律分析 Skill

专业的裁判文书生成与质量评判辅助工具，基于最高人民法院权威文献，提供完整的裁判思维体系、方法论体系和评判标准体系

Nomos — a programming language for legal reasoning. Typed rules with jurisdiction and validity dates, LLM-powered fact extraction, defeasible logic, proof trees that cite statutes and cases. Apache-2.0. Experimental.

The OSCAR Project: the general-purpose defeasible reasoner and architecture for a rational agent

类案办案要件指南智能辅助系统 - 基于AI的案件类型识别与要件式审判清单生成工具

Curry-Howard isomorphism for deontic logic — formalizing Korean Civil Act with Haskell type-level stratified defeasibility

Adaptive Collaboration of Arena-Based Argumentative LLMs for Explainable and Contestable Legal Reasoning

CAP RLVR: Reinforcement Learning from Human Feedback for Legal Reasoning using Caselaw Access Project data. Complete GRPO training pipeline with OpenAI Gym environments, deterministic reward functions, and multi-stage curriculum learning for legal LLM development.

Open-source reasoning kernel for auditable EU AI Act compliance classification using t-norm fuzzy logic

Graph-constrained legal reasoning for Indian judicial AI · FalkorDB · IRAC schema · arXiv:2605.14665

Judicial accountable record system for opinions, claims, holdings, citations, dependencies, and later treatment.

PRC-first legal semantic alignment framework for constraining US legal concepts within Chinese-law reasoning contexts.

HECTOR (Hierarchical Evaluation of Civil-Criminal Textual Orchestrator & Retrieval) is a zero-hallucination, Hard-RAG legal intelligence system for Indian law, specializing in IPC to BNS mapping with precise citations.

A structured and rule-guided multi-agent framework for interpretable legal reasoning.

How legal values decay when AI inherits precedents without context — Synthetic Legalism research

LegalEval, high level framework for evaluation of legal LLMs and reasoning capabilities.

A fact-centered legal narrative generation system that uses Answer Set Programming (ASP) to guide Large Language Models in creating structured legal narratives for fee waiver applications.

JACV: A Benchmark for Legal-Argumentative Coherence Verification in Brazilian Portuguese

LegalScope benchmark for measuring exam-to-case transfer in LLM legal reasoning across public exams and de-identified Chinese civil judgments.

Tell coherence from truth — an LLM prompt that lays out any argument's hidden premises, alternatives, and missing evidence