Claude for Legal: Anthropic's AI-Powered Legal Workflow Plugins — Complete Guide 2026

What Is Claude for Legal?

Claude for Legal is an open-source suite of AI plugins released by Anthropic that turns Claude into a specialized legal workflow assistant. Hosted on GitHub (anthropics/claude-for-legal), it contains 60+ reference agents, skills, and data connectors spanning every major practice area — from commercial contracts and privacy compliance to litigation, IP prosecution, corporate M&A, employment law, AI governance, and even law school clinics.

What makes it different from a generic AI chatbot is its depth. Each agent comes with a cold-start interview that learns your firm's playbook, a practice-specific CLAUDE.md profile that every skill reads from, source-attributed citations, jurisdiction awareness, and explicit guardrails that keep the lawyer in the driver's seat. It's not a replacement for legal judgment — it's a force multiplier for the work lawyers do every day.

The repo is structured around three delivery mechanisms: Claude Cowork plugins (for legal professionals using Claude's collaborative workspace), Claude Code plugins (for developer-oriented workflows), and Managed Agents API (for firms that want to embed these agents into their own workflow engines). Same system prompts, same skills — you choose where it runs.

Available Practice-Area Plugins

Commercial Legal

Eight plug-ins for the in-house commercial team:

Vendor Agreement Reviewer — Reviews vendor MSAs against your playbook and produces a redline memo
NDA Triager — GREEN/YELLOW/RED triage of inbound NDAs so only the hard ones hit a lawyer's desk
Amendment Tracer — Traces how a contract has changed across its base agreement and every amendment
Renewal Watcher — Scans the contract register for cancel-by and renewal deadlines (scheduled agent)
Deal Debrief — Weekly sweep of signed agreements with playbook deviations
Playbook Monitor — Watches the deviation log and proposes playbook updates
Escalation Router — Routes contract issues to the right approver and drafts the ask
Tabular Diligence Review — Tabular review over a data room with one row per document

Privacy Legal

Seven agents covering the privacy workflow from intake through monitoring:

DSAR Responder — Drafts DSAR acknowledgments and substantive responses within statutory timelines
DPA Reviewer — Reviews a DPA against your playbook as controller or processor
PIA Generator — Generates a Privacy Impact Assessment in house format
Privacy Triager — Decides whether a processing activity needs a PIA, GDPR DPIA, or can proceed
Privacy Reg Gap Checker — Diffs new or changed regulation against current privacy policy
Privacy Policy Monitor — Sweeps saved PIAs, DPA reviews, and triage results for policy drift
Launch Reviewer — Reviews product launches against your risk calibration

Corporate Legal

For M&A, entity management, and integration teams:

Issue Extractor — Reads VDR documents and extracts issues per house categories
Board Consent Drafter — Drafts unanimous written consents with precedent search
Material Contracts Schedule Builder — Builds disclosure schedules from diligence findings
Entity Compliance Tracker — Computes filing deadlines across jurisdictions
Closing Checklist Driver — Tracks every condition, consent, and document blocking close
Integration Runbook — Phased post-closing integration plan with consent tracking
Data Room Watcher — Monitors the VDR for new uploads (scheduled agent)

Employment Legal

Eight agents for the full employment lifecycle:

Termination Reviewer — Runs proposed terminations against jurisdiction-specific risk flags
Hire Reviewer — Reviews offer letters and restrictive covenants
Worker Classification Screener — Tests engagements against the controlling state test
Leave Tracker — Monitors open leaves with FMLA/CFRA/PFL/ADA deadlines
Investigation Lead — Opens, tracks, and summarizes internal investigations
Policy Drafter — Drafts employment policies with state supplements
International Expansion Planner — Kicks off EOR-vs-entity planning for new countries
Wage & Hour Q&A — Jurisdiction-aware Q&A for quick questions

Litigation Legal

Thirteen agents for the full litigation lifecycle:

Claim Chart Builder — Element-by-element claim charts for patents or civil causes
Docket Watcher — Monitors court dockets for filings and deadlines
Demand Letter Drafter — Drafts demands with FRE 408 awareness
Demand Intake/Received Triage — Structured context gathering and inbound triage
Subpoena Triage — Classifies and plans compliance with new subpoenas
Chronology Builder — Builds or updates chronologies from declared sources
Deposition Prep — Deposition outlines tied to case theory
Brief Section Drafter — Drafts brief sections in house style
Privilege Log Reviewer — First-pass privilege log review
Legal Hold — Issue, refresh, release, or report on holds
Matter Intake & Briefing — Uniform intake and deep briefing for GC calls
Portfolio Status & OC Status — Risk distribution and outside counsel updates

Intellectual Property

Trademark Clearance Screener — First-pass knockout check
Cease & Desist Drafter — Calibrated to your enforcement posture
DMCA Takedown — Drafts takedowns and counter-notices
OSS Compliance Checker — License classification for your deployment model
FTO Triager — Structured first look at potentially blocking patents
Infringement Triager — Multi-factor triage across TM/copyright/patent/trade secret
IP Clause Reviewer — Reviews assignment, license grants, warranties
IP Portfolio Tracker — Registrations, renewals, maintenance fees
IP Renewal Watcher — Scheduled deadline reports (scheduled agent)

AI Governance

AI Use Case Triager — Classifies AI use cases against your registry
AI Impact Assessor — Runs AIAs across relevant regimes
Vendor AI Reviewer — Reviews vendor AI terms for training-data and liability
AI Reg Gap Checker — Diffs new AI regulations against your posture
AI Policy Monitor — Sweeps for AI-policy drift (scheduled agent)

Regulatory & Product Legal

Reg Feed Watcher — Polls regulatory feeds, writes the Monday digest
Policy Diff & Gap Tracker — Diffs changes and tracks open gaps
Policy Redrafter — Proposed redrafts for policy owner review
NPRM Comment Tracker — Tracks comment periods and deadlines
Marketing Claims Checker — Flags copy needing substantiation
"Is This a Problem?" Triage — Fast Slack-question pattern matching
Launch Watcher — Watches launch tracker for upcoming reviews (scheduled agent)

Law School & Legal Clinic

Clinic Intake — Structured client intake with conflict flags
Case Memo Scaffold — IRAC-scaffolded analysis memos
Research Roadmap — Statutes, case law areas, search terms
Bar Prep Coach — MBE and essay practice targeted at weak subjects
Socratic Drill Sergeant — Interactive Q&A drill (does not give answers)
IRAC Grader — Grades essays on structure, issue-spotting, and analysis
Case Briefer & Outline Builder — Brief cases and build outlines from notes

How It Works: Three Deployment Options

Claude for Legal agents are available through three channels, all sharing the same source code:

1. Claude Cowork Plugins

For legal professionals working in Claude's collaborative workspace. Install via the marketplace in 60 seconds — described in the QUICKSTART.md. Each plugin presents a cold-start interview that learns your firm's templates, jurisdiction preferences, privilege protocols, and risk tolerance. After setup, you invoke agents via slash commands like /commercial-legal:review or /privacy-legal:dsar-response.

2. Claude Code Plugins

For legal tech teams and developer-oriented workflows. Same agents, same prompts, but running in a terminal environment that integrates with version-controlled playbooks and CI/CD pipelines.

3. Claude Managed Agents API

For law firms and legal departments that want to embed these agents into their own workflow engines. Deploy the same agents behind custom interfaces, integrate with your document management system (iManage, NetDocuments), and run scheduled agents (Renewal Watcher, Docket Watcher, Reg Feed Watcher) as background services.

Key Features That Matter for Legal Work

Source Attribution on Every Citation

Every factual claim, case reference, and regulatory cite includes a source link. No mysterious "the law says" — you can check the underlying authority before relying on the output.

Privileged Document Handling

Agents are designed with conservative defaults around privilege. They flag documents that may contain attorney-client privileged or work-product protected material, and they surface jurisdiction-specific privilege rules.

Authority Levels

Each agent operates within defined authority levels: screening and summarization (no legal conclusion), flagged recommendation (needs attorney review), and draft generation (attorney must verify before acting). This maps to the ABA's guidance on AI use in legal practice.

Jurisdiction Awareness

Employment agents know the difference between California and Texas employment law. Privacy agents distinguish GDPR from CCPA from LGPD. Corporate agents compute filing deadlines across Delaware, California, and foreign jurisdictions.

Cold-Start Interview & Practice Profiles

Every plugin begins with a CLAUDE.md interview that captures your firm's templates, risk tolerance, preferred outside counsel, privilege logging practices, and filing deadlines. This profile persists and informs every downstream skill.

Scheduled and Watcher Agents

Several agents run in the background: Renewal Watcher scans contract registers, Docket Watcher monitors PACER filings, Reg Feed Watcher polls regulatory feeds. These deliver proactive alerts rather than waiting for a lawyer to ask.

Comparison: Claude for Legal vs CoCounsel vs Harvey AI

CoCounsel (Thomson Reuters / Casetext)

CoCounsel is the most established legal AI platform, powered by GPT-4 and trained on Westlaw's authoritative legal database. It excels at legal research, document review, and deposition preparation. Its strengths are deep integration with Westlaw and a mature commercial product. Its limitations: it's a closed platform (no plugin ecosystem), priced per-seat at enterprise tiers, and its research is Westlaw-dependent. Claude for Legal's advantage is the open-source, plugin-based architecture — you can customize agents for your specific practice, connect to any document system, and build workflows that CoCounsel's closed model cannot support.

Harvey AI

Harvey AI (backed by OpenAI) focuses on large law firm workflows with deep integrations into firm document management systems. It's built on custom-trained models and has strong brand recognition in Am Law 100 firms. Its limitations: it's an enterprise-only product with significant per-user costs, limited transparency about its training data, and a walled-garden approach to integrations. Claude for Legal offers comparable depth at no licensing cost (open-source), full transparency about prompts and system design, and the flexibility to run on your infrastructure or via Anthropic's API.

When to Choose Which

Feature	Claude for Legal	CoCounsel	Harvey AI
Pricing Model	Open-source (free)	Enterprise seat-license	Enterprise seat-license
Customization	Full (open plugins)	Limited	Limited
Legal Research	General web + cited sources	Westlaw-integrated	Custom corpus
Plug-in Ecosystem	60+ agents, extensible	Fixed workflows	Fixed workflows
Self-Hosted	Yes (Managed Agents API)	No	No
Transparency	Full (open-source prompts)	Limited	Limited
Scheduled Agents	Yes	No	Limited

Practical Use Cases

For Law Firms

Streamline NDA review — The NDA Triager reduces the 80% of routine NDAs to a GREEN/YELLOW/RED flag, letting associates focus on the complex ones
Automate contract diligence — In M&A, the Tabular Diligence Review and Issue Extractor process data rooms at document-by-document granularity
Litigation workflow automation — From intake through discovery, deposition prep, and brief drafting, thirteen agents cover the full lifecycle
IP portfolio management — Track registrations, renewals, and maintenance fees across jurisdictions with the IP Portfolio Tracker and Watcher

For In-House Counsel

Privacy compliance at scale — DSAR Responder, DPA Reviewer, and PIA Generator handle the compliance workflow while you focus on strategic privacy decisions
Product launches without bottlenecks — The Launch Reviewer, Marketing Claims Checker, and "Is This a Problem?" Triage keep legal review from becoming the critical path
AI governance readiness — Five dedicated agents cover AI use case triage, impact assessment, vendor review, regulatory gap analysis, and policy monitoring
Employment law across states — Termination reviews, hiring docs, worker classification, and leave tracking with jurisdiction-specific rules

For Law Schools & Clinics

Student practice with guardrails — Clinic Intake, Case Memo Scaffold, and IRAC Grader help students learn in a structured environment
Bar prep — The Bar Prep Coach and Socratic Drill Sergeant (which refuses to give answers, forcing true Socratic learning) are novel AI applications for legal education

Important Limitations

Anthropic is explicit about what Claude for Legal is not:

Not legal advice. Every output is a draft for attorney review. The plugin is an aid, not a substitute for professional judgment.
Not a legal conclusion. Skills include checklists, frameworks, and case characterizations as aids to attorney analysis — not statements of law.
Draft-for-review model. A lawyer reviews, verifies, and takes professional responsibility for anything that leaves the building.
Guardrails built in. Source attribution on every citation, conservative defaults on privilege, jurisdiction assumptions surfaced, and explicit gates before anything is filed or sent.
Not Anthropic's legal positions. The tools help lawyers analyze issues; they don't represent Anthropic's view of the law. Many areas covered are unsettled and evolving.

The repo's README puts it bluntly: "Every output from these plugins is a draft for attorney review — not legal advice, not a legal conclusion, not a substitute for a lawyer."

Getting Started: 60-Second Install

Getting started with Claude for Legal is straightforward:

Install Claude Desktop — Download from claude.com/download
Get Claude Cowork access — Required for the plugin marketplace
Visit the GitHub repo — QUICKSTART.md has installation instructions and a video walkthrough
Choose your plugins — Browse the agents listed above and install the ones that match your practice areas
Run the cold-start interview — Each plugin's CLAUDE.md will walk you through setting up your practice profile, templates, and preferences
Start working — Use slash commands like /commercial-legal:review or /privacy-legal:dsar-response to invoke agents

For firms that want deeper integration, the Claude Managed Agents API allows deploying the same agents behind your own interfaces, connecting them to your DMS, and running scheduled background agents. Check the repo's README for managed-agent cookbooks covering renewal watchers, docket monitors, regulatory feed monitors, diligence grids, and launch radars.

MCP Connectors for Legal Systems

The repo includes MCP (Model Context Protocol) connectors for both general productivity tools (Slack, Google Drive, Box) and legal-specific systems (Ironclad, DocuSign, iManage, Everlaw, CourtListener, and more). These connectors let agents read documents from your DMS, file documents via e-filing systems, check court dockets, and post to Slack — all within the workflow.

Bottom Line

Claude for Legal is the most comprehensive open-source legal AI suite available in 2026. With 60+ specialized agents across 10 practice areas, source-attributed citations, jurisdiction-aware workflows, and a clear "draft for attorney review" model, it gives law firms and legal departments a powerful, transparent, and customizable alternative to closed platforms like CoCounsel and Harvey AI.

Is it a replacement for lawyers? Absolutely not. But for the workflows that consume hours of attorney time — contract review, compliance checks, litigation prep, regulatory monitoring — it's a force multiplier that lets lawyers focus on the judgment calls that only a human can make.