Discover and explore top open-source AI tools and projects—updated daily.
evolsbAI contract analysis agent skill
Top 94.1% on SourcePulse
Summary This project offers an AI-powered contract review skill for integration with coding assistants like Claude Code and Codex. It addresses the lack of accessible, developer-friendly legal AI tools by providing CUAD-based risk detection, market benchmarks, and actionable redlines, enabling contract analysis directly within development workflows. The skill benefits individual developers and legal professionals seeking efficient, nuanced contract analysis without enterprise costs or complex integrations.
How It Works
The skill integrates as a component for Agent Skills-compatible tools, using the CUAD dataset (41 risk categories) for analysis. Its core innovation is position-aware review, adapting flags and suggestions based on the user's specified party role. It provides document-type specific checklists, market standard benchmarks with clear thresholds, and concrete redline language suggestions, moving beyond generic advice.
Quick Start & Requirements
Installation involves cloning the repository into the AI coding assistant's skills directory (e.g., git clone https://github.com/evolsb/claude-legal-skill ~/.claude/skills/contract-review). Compatibility with the Agent Skills standard is required. Generating Word/PDF redlines necessitates installing legal-redline-tools via pip. No specific hardware or OS prerequisites beyond tool compatibility are listed.
Highlighted Details
legal-redline-tools for Word/PDF generation.Maintenance & Community
Maintained by evolsb. Issues can be reported via GitHub. Direct contact is available at chris@ctsheehan.com. No specific community channels or roadmap details are provided.
Licensing & Compatibility
Released under the MIT License. Analysis defaults to US law; explicitly not legal advice and requires review by qualified counsel for material terms.
Limitations & Caveats
Not a substitute for professional legal counsel; focuses on US law, with significant jurisdictional variations. Very long contracts may require section-by-section review due to context window limits. ContractEval benchmarks show ~0.62 F1 for clause extraction, best for initial review and issue flagging, not definitive legal judgment.
2 months ago
Inactive