Anthropic releases Claude for Legal

Anthropic has released Claude for Legal on GitHub, a suite of eleven legal plugins and MCP connectors deploying via Claude Cowork, Claude Code, or APIs.

Anthropic makes available on GitHub the Claude for Legal repository, under an Apache 2.0 license, which includes plugins, skills, agents, and MCP connectors for eleven legal domains: commercial, corporate, litigation, IP, privacy, regulatory compliance, AI governance, employment law, product law, plus two training modules (legal-clinic and law-student). The suite installs into Claude Cowork or Claude Code, or deploys via the Claude Managed Agents API.

Each plugin starts with a cold-start interview that collects the team's jurisdictions, playbook, and escalation rules to generate a practice profile (CLAUDE.md) subsequently read by all skills. The default calibration remains very American (FMLA, FRE 408, DMCA, MBE, CourtListener, and Westlaw integrations), but the system is explicitly multi-jurisdictional: the privacy-legal module covers GDPR DPIA, employment-legal integrates an International Expansion Planner, and regulatory-legal applies its gap analysis to any configured regime, including the AI Act. A legal-builder-hub manages the installation of community skills under a trust layer, and a partner plugin from Thomson Reuters, CoCounsel Legal, adds Westlaw Deep Research with cited reports. Anthropic reminds that each output remains a draft intended for attorney review, with source attribution and safeguards before any irreversible action.