Agent♥︎Age
Catalog

Rechtssysteem.ai

Official

by rechtssysteem-ai · Python

Lekkage-vrije voorspelling van Nederlandse rechtszaken, getraind op 609.715 uitspraken

AI.Rechtssysteem / Rechtssysteem MCP

This MCP server provides AI agents access to “lekkage-vrije” prediction of outcomes for Dutch legal cases. It is trained on 609,715 judgments and is designed to evaluate predictions on text with answer-revealing sentences removed (knipregel R2). Reported metrics include 78,2% accuracy and 77,1% Macro-F1.

🛠️ Key Features

  • Leakage-free prediction approach for Dutch court cases
  • Trained on 609.715 cases (uitspraken)
  • Answer leakage reduction via sentence removal (knipregel R2)
  • Reported restlekkage of 0,1%

🚀 Use Cases

  • Predicting outcomes of Nederlandse rechtszaken for AI agents
  • Benchmarking legal-tech models using leakage-controlled evaluation

⚡ Developer Benefits

  • Clear dataset coverage: 609.715 zaken
  • Published evaluation metrics: Accuracy 78,2% and Macro-F1 77,1%
  • Topic alignment for MCP and legal-tech workflows: benchmark, mcp, model-context-protocol, rechtspraak, legal-tech

⚠️ Limitations

  • No additional tool list or MCP tool behavior is provided in the available source excerpt.

Topics

benchmarklegal-techmcpmodel-context-protocolrechtspraak