anthropics/claude-for-legal

What it solves

This project provides a comprehensive set of AI agents, skills, and data connectors designed to automate and assist with common legal workflows. It aims to reduce the manual effort involved in contract review, regulatory monitoring, and legal research while ensuring that a human attorney remains the final reviewer and decision-maker.

How it works

The system is built as a collection of plugins that can be installed via Claude Cowork or Claude Code, or deployed as Managed Agents via an API. It uses a combination of system prompts, practice-specific profiles (CLAUDE.md), and MCP connectors to integrate with productivity tools (Slack, Google Drive) and legal-specific software (Ironclad, DocuSign, iManage).

Each plugin is tailored to a specific legal practice area (e.g., commercial, corporate, privacy, IP, litigation) and often begins with a "cold-start interview" to learn the user's specific legal playbook.

Who it’s for

• In-house counsel managing commercial agreements, privacy, and product launches.
• Law firms handling M&A diligence, litigation, and IP portfolios.
• Law students and clinic participants needing help with case memos, IRAC grading, and bar prep.
• Legal engineers building custom legal AI workflows.

Highlights

• Practice-Area Specialization: Dedicated plugins for 11+ legal domains including AI governance, employment, and regulatory law.
• Workflow Agents: Pre-built agents for specific tasks like Vendor Agreement Review, DSAR Responders, and Claim Chart Builders.
• Scheduled Automation: "Cookbooks" for agents that monitor feeds, such as renewal watchers and docket watchers.
• Legal Guardrails: Built-in source attribution and explicit gates to ensure outputs are drafts for attorney review, not legal advice.