A suite of plugins for legal workflows
8.7k
Stars
1.6k
Forks
58
Open issues
6
Contributors
AI Analysis
Claude for Legal is a suite of specialized AI plugins and agents designed for legal professionals across multiple practice areas—in-house counsel, law firms, law schools, and regulatory teams. It provides domain-specific workflows for contract review, due diligence, compliance monitoring, and litigation support via Claude Cowork plugins, Claude Code, or the Managed Agents API. This tool is built specifically for lawyers and legal departments; it is not a general-purpose AI assistant and expli...
Inferred from signals mentioned in the README (tests, CI, type safety) — not a review of the actual code.
AI's overall editorial judgment — not an average of the bars above, can weigh other factors too.
Anthropic releases Claude plugin suite targeting the full stack of legal workflows from contract review to regulatory monitoring
Claude for Legal is a collection of practice-area plugins, named workflow agents, MCP connectors, and managed-agent cookbooks built on top of Claude. It targets in-house legal teams, law firms, and law school clinics across commercial, employment, corporate, privacy, IP, litigation, and regulatory work. The project ships two deployment modes: a Claude Cowork/Claude Code plugin path for individual attorneys and a Managed Agents API path for teams embedding it in their own workflow infrastructure. Every output is framed as a draft for attorney review, with built-in guardrails for privilege, jurisdiction disclosure, and citation sourcing.
Created in April 2026, likely as part of Anthropic's push to build vertical-specific plugin ecosystems alongside the launch of Claude Cowork and the Managed Agents API. It is a first-party reference implementation, not a community project.
8,486 stars accumulated in roughly two months points to strong initial interest, likely driven by Anthropic's own announcement channels and the broader legal-tech community's appetite for AI-assisted workflows. The 111 stars gained in the last 7 days suggests the initial launch spike has settled into a slower but still active growth curve. The 1,590 forks indicate meaningful hands-on experimentation, not just passive watching.
Adoption not verified from external sources. The project is first-party from Anthropic, which increases the probability of real internal usage and coordinated enterprise pilots, but no documented production deployments, case studies, or external testimonials are visible in the README or metadata. The high fork count (1,590) suggests practitioners are adapting it, but whether those forks represent production use or experimentation is unclear.
Appears to follow a plugin-per-practice-area structure, each plugin exposing named skills and a shared CLAUDE.md practice profile that agents read at runtime. Likely uses a cold-start interview pattern to capture team-specific playbook parameters before running any workflow. MCP connectors appear to be separate modules targeting both general productivity tools (Slack, Google Drive, Box) and legal-specific systems (Ironclad, iManage, Everlaw, CourtListener). Managed-agent cookbooks appear to be separately runnable scheduled-agent definitions rather than interactive plugins.
Not documented in README.
Last push was June 21, 2026 — four days before the evaluation date — indicating active, ongoing development. The repo is less than three months old and appears to be receiving regular commits. No signs of abandonment or stagnation based on available metadata.
ADOPT IF: your team is already on Claude Cowork or the Managed Agents API, has licensed attorneys who will own review of every output, and needs to accelerate contract review, diligence, or regulatory monitoring workflows without building prompt infrastructure from scratch. AVOID IF: you need a fully integrated, audited, enterprise-grade legal platform with SLAs, compliance certifications, and a vendor support contract — this is a reference implementation, not a managed product. MONITOR IF: you are evaluating Claude-based legal tooling but need more evidence of production reliability, jurisdiction accuracy, and privilege-handling correctness before committing to a deployment.
Independent dimensions
Mainstream potential
5/10
Technical importance
7/10
Adoption evidence
2/10
- Legal accuracy risk: jurisdiction-specific outputs and case law citations are generated by a language model; errors in a legal context carry professional liability consequences that most software risks do not.
- Dependency risk: the entire plugin suite is tightly coupled to Anthropic's API, Claude Cowork, and the Managed Agents API — pricing changes, model deprecations, or API breaking changes would require significant rework.
- Maturity risk: the repo is under three months old with no documented production deployments; early adopters are effectively beta testers in a high-stakes domain.
- Scope creep risk: covering ten-plus practice areas in a single repo risks shallow coverage in any one area; teams with deep specialization (e.g., patent prosecution, complex litigation) may find the agents insufficiently tailored.
- Regulatory risk: AI-generated legal outputs are under active regulatory scrutiny in multiple jurisdictions; court rules on AI disclosure and bar ethics opinions on AI-assisted practice are still evolving and could constrain how these tools are used.
Likely to become the reference starting point for Claude-based legal automation, with enterprise adoption growing as Anthropic matures its Cowork and Managed Agents platform — but will probably remain a scaffold to customize rather than a deploy-as-is product.
Newsletter
Get analyses like this every Monday
Free weekly digest of the most interesting open-source discoveries.
Languages
Information
- Language
- Python
- License
- Apache-2.0
- Last updated
- 3d ago
- Created
- 3mo ago
- Analyzed with
- anthropic/claude-haiku-4-5
Stars over time
Contributors over time
Top 100 contributors only — repos with more will plateau at 100.
Open issues
Feature request: Local pre-processing step before Cowork ingests large local file sets (esp. for legal/document-heavy workflows)
Custom config doesn't persist in Claude Cowork
Connector design question: is a verifiable-attestation identifier a first-class citation source for ai-governance-legal?
Top contributors
Recent releases
No releases published yet.
Similar repos
anthropics/financial-services
Claude for Financial Services provides reference agents, skills, and data...
anthropics/knowledge-work-plugins
Knowledge Work Plugins is a collection of 11 open-source Claude plugins...
CloudAI-X/claude-workflow-v2
A Claude Code workflow plugin that integrates specialized agents, skills,...
anthropics/claude-plugins-official
An official Anthropic-managed marketplace and directory for Claude Code...
| Repository | Stars | Week Δ | Language | Score | Updated |
|---|---|---|---|---|---|
|
anthropics/claude-for-legal
this
|
8.7k | +99 | Python | 8/10 | 3d ago |
|
anthropics/financial-services
|
33.3k | — | Python | 8/10 | 2w ago |
|
anthropics/knowledge-work-plugins
|
22.5k | — | Python | 8/10 | 12h ago |
|
CloudAI-X/claude-workflow-v2
|
1.4k | — | Python | 7/10 | 2w ago |
|
anthropics/claude-plugins-official
|
31.9k | — | Python | 8/10 | 7h ago |
|
anthropics/claude-agent-sdk-python
|
7.6k | — | Python | 8/10 | 9h ago |
Harvey is a closed, venture-backed legal AI product with a dedicated sales motion and enterprise contracts. Claude for Legal is open-source, self-hostable, and composable with existing infrastructure. Harvey has documented law firm adoption at scale; Claude for Legal's production adoption is unverified but may appeal to teams that need customization or want to avoid vendor lock-in.
Ironclad AI focuses narrowly on contract lifecycle management with deep CRM and workflow integrations. Claude for Legal is broader in scope (employment, litigation, regulatory, IP) but less deeply integrated into contract execution workflows out of the box, though it does include an Ironclad MCP connector.
The general Claude plugins repo is the parent ecosystem; Claude for Legal is a vertical specialization within it. Teams needing general-purpose agent patterns would use the official plugins; teams needing legal-specific guardrails, privilege handling, and jurisdiction-aware outputs would use this repo.
Knowledge-work plugins serve a broader white-collar productivity market. Claude for Legal trades breadth for depth in legal-specific compliance concerns (privilege, jurisdiction, citation sourcing, filing gates) that a generic knowledge-work plugin would not address.
A community-built Claude workflow library with much lower adoption and no apparent legal specialization. Claude for Legal benefits from first-party status and more thorough domain coverage, though community repos sometimes move faster on niche integrations.