name: ContractBench
description: >
  ContractBench measures how well an AI agent reviews and redlines a commercial contract the way an in-house attorney
  would: deciding where to intervene, what to change, what to leave alone, when to ask the counterparty a question and
  when to escalate a point to its own side. Each task is a real Word document with grounding documents and a brief; the
  agent works in the .docx with tracked changes and comments and is graded against pass-or-fail rubric items written by
  the attorney who reviewed the same contract. 350 tasks from 35 scenarios (210 public, 140 held out), five attorneys per
  scenario, two negotiation turns. Scores are the weighted rubric pass rate on completed rollouts, 0 to 100;
  leaderboard.md in this repository lists the published results.
evaluation_framework: harbor

tasks:
  - id: contractbench_private
    config: default
    split: test
  - id: contractbench_public
    config: default
    split: test
  - id: contractbench_full
    config: default
    split: test