Anthropic Open-Sources Claude for Legal: Plugin Suite for Contract Review, NDA Triage, and More
Anthropic has open-sourced Claude for Legal, a suite of plugins, managed-agent cookbooks, and MCP connectors designed for in-house, firm, and academic legal work. The repo is structured around practice areas: commercial, privacy, product, corporate, employment, litigation, regulatory, AI governance, IP, and law student clinics.
Key Agents and Commands
Each agent maps to a specific workflow and is invoked via a plugin command. Notable agents include:
- Vendor Agreement Reviewer —
/commercial-legal:review: Reviews vendor MSAs against your playbook and produces a redline memo. - NDA Triager —
/commercial-legal:review: GREEN/YELLOW/RED triage so only hard NDAs hit a lawyer's desk. - Amendment Tracer —
/commercial-legal:amendment-history: Traces contract changes across amendments. - Renewal Watcher — scheduled agent that scans for cancel-by and renewal deadlines.
- Deal Debrief — weekly sweep of signed agreements with playbook deviations.
Agents cover litigation, corporate, employment, privacy, regulatory, AI governance, IP, and law student use cases.
Architecture and Deployment
The system prompt and skills are shared across two deployment options:
- Claude Cowork / Claude Code plugin — install via marketplace instructions in QUICKSTART.md (60-second install).
- Claude Managed Agents API — deploy behind your own workflow engine.
Each plugin runs a cold-start interview that learns your playbook and a CLAUDE.md practice profile that every skill reads.
MCP Connectors
Connectors include general productivity tools (Slack, Google Drive, Box) and legal-specific systems:
- Ironclad
- DocuSign
- iManage
- Everlaw
- CourtListener
Guardrails and Disclaimers
Every output is treated as a draft for attorney review — not legal advice. Features include: source attribution on every citation, conservative defaults on privilege and subjective legal calls, jurisdiction assumptions surfaced, and explicit gates before anything is filed or sent. The repo explicitly states: "These plugins do not represent Anthropic's legal positions."
Getting Started
The README points to QUICKSTART.md for a 60-second install. The repo is open-source (starred 4.4k, forked 641) and accepts contributions via the CONTRIBUTING.md guidelines.
📖 Read the full source: HN AI Agents
👀 See Also
Nelson: A Claude Code Plugin for Coordinating AI Agents Like a Naval Fleet
Nelson is a Claude Code plugin that structures AI agent coordination using naval fleet principles, featuring three execution modes, a risk classification system, hull integrity monitoring, and standing order gates to prevent common anti-patterns.
Efficient Token Management with Open-Source MCP Servers: Pare
Pare MCP servers reduce token waste and enhance efficiency when AI coding agents use developer tools by providing structured output.
Head-to-head code review experiment compares three AI tools on same codebase
A video experiment tests Codex, Claude Code, and Claude Code with Sextant on identical code review tasks, with Codex verifying findings and judging which report is more valuable. The focus is on how workflow and structure affect what AI notices and prioritizes.
Lightpanda: Open-source headless browser for LLM agents with native MCP server and markdown output
Lightpanda is an open-source headless browser designed for LLM-powered agents that uses 16x less memory than Chrome (215MB vs 2GB) and completes web crawling benchmarks in 5 seconds instead of 47 seconds. It provides native markdown output, semantic tree with interactivity detection, and a built-in MCP server.