Claude for Legal (GitHub repository)

The Claude for Legal repository is a bundle of reference agents, skills, and data connectors for legal workflows, covering in-house commercial, privacy, product, corporate, employment, litigation, regulatory, AI governance, IP, and even law school clinics and students. The same system prompts and skills can be installed as a Claude Cowork or Claude Code plugin, or deployed behind your own workflow engine via the Claude Managed Agents API.

The core structure branches into three parts.

• Practice-area plugins: learn your team's playbook through an initial interview and use a CLAUDE.md profile that all skills reference.
• Managed-agent cookbooks: handle recurring check-type tasks such as renewal watcher, docket watcher, regulatory feed monitor, diligence grid, and launch radar.
• MCP connectors: connect common tools like Slack, Google Drive, and Box as well as legal systems like Ironclad, DocuSign, iManage, Everlaw, and CourtListener.

The named agents are broken down into concrete task units such as Vendor Agreement Reviewer, DSAR Responder, Termination Reviewer, Launch Reviewer, and AI Impact Assessor. Each agent produces drafts and decision support based on documents, tying together workflows from contract review to regulatory response and AI governance into a single command.

All outputs are designed as drafts for attorney review. They include source citations, conservative privilege defaults, explicit statements of jurisdictional assumptions, and gates placed before filing, sending, or relying on outputs, aiming to speed up review rather than deliver legal conclusions, and the tool does not represent Anthropic's legal position. Final responsibility rests with the attorney using it.