jjgujkk/LegalMind-Agent-

LegalMind is an AI-agent driven enterprise contract-review platform for legal, finance and business teams. It performs parallel, multi-agent analysis with long-chain reasoning to auto-identify high-risk clauses, cross-clause conflicts, cross-border compliance gaps and produces executable redlines and negotiation strategies—reducing review time (typ. ~8 minutes for a 30–50 page contract) and improving defect detection. Core architecture: a master controller that semantically chunks documents and dynamically schedules four parallel sub-agents (risk ID, clause compliance, financial risk, cross‑border regulation), plus a result aggregator with confidence scoring and an interactive clarification loop. Key features: multi-format ingest (PDF/DOCX/TXT), cross-jurisdiction rule libraries (GDPR, CCPA, China personal data law, export controls), contradiction and latent‑obligation detection, negotiation priority ranking, clause‑level visual heatmap and four‑dimensional annotation (legal effect, financial impact, operational cost, negotiation weight), one‑click clean redlines, and plain‑language summaries for non‑lawyers. Tech stack highlights: LangGraph/LangChain orchestration, Qwen-72B (swapable LLMs), Neo4j knowledge graph, Qdrant vector DB, FastAPI backend, Streamlit/React front end, unstructured.io and pdfplumber for parsing, and custom CoT prompt chains with confidence thresholds.

このRepoのdescription / GitHub topics / READMEのAI要約に出現したAI関連キーワードをカテゴリ別に表示。各バッジは該当カテゴリの詳細ランキングへリンクします。