See how Spellbook helps legal teams draft contracts faster than ever before.
Book Demo
The Claude legal plugin is a free, open-source plugin that turns Claude into a structured contract review tool for in-house legal teams and law firms. Anthropic ships 12 such plugins under Claude for Legal, covering commercial, corporate, privacy, employment, intellectual property, litigation and AI governance work, and each one runs in Claude Cowork, Claude Code or the Claude for Microsoft 365 sidebar on a paid Claude plan from $17 per month. The plugin costs nothing; the spend sits in that plan and in the systems it connects to. Every output arrives as a draft for attorney review.
The January 2026 version was a single legal plugin with five generic commands; the May 2026 suite is 12 plugins, about 150 named skills across them, and a set of connectors, and the contract skills now return tracked changes in Word rather than text in a chat window. Teams pricing it today, in-house legal teams especially, are weighing the per-plugin setup cost against what the suite does once it is configured.
The Claude legal plugin is one plugin in Claude for Legal, the open-source suite Anthropic published on GitHub in May 2026 for legal work. The suite holds 12 practice-area plugins, about 150 named skills across them, and Model Context Protocol connectors (an open standard that lets Claude query a system directly instead of working from an uploaded copy) into Ironclad, DocuSign, iManage, Everlaw, CourtListener and Westlaw through CoCounsel Legal. It replaces the single Cowork legal plugin Anthropic shipped in January 2026.
Before a skill produces anything useful, each plugin runs a cold-start interview, ten to twenty minutes of questions that build a `CLAUDE.md` practice profile from the team’s own playbook, a signed template and a prior review memo. Every skill in the plugin reads from that profile rather than from general training knowledge. Teams curious about how lawyers use Claude day to day outside this specific suite will recognize the same chat-based workflow underneath, pointed now at a practice profile instead of a blank conversation. Teams can reach the same skills two ways:
The plugin reviews vendor agreements, non-disclosure agreements (NDAs) and software-as-a-service (SaaS) subscriptions against the practice profile, flags playbook deviations, drafts clause language, and answers data subject access requests on a statutory clock. Because every output remains a draft, whether an attorney may rely on it for client work is a question of professional-responsibility rules, not of the plugin itself. Its real limit is the practice profile: a thin cold-start interview produces generic output, and without a connected research tool every citation the plugin returns is flagged `[verify]`.
[cta-1]
| Plugin | What it covers |
|---|---|
| commercial-legal | Vendor agreements, NDAs, SaaS subscriptions and contract triage for commercial teams |
| corporate-legal | Entity compliance, cap-table housekeeping and tabular multi-document review |
| privacy-legal | DSAR response, DPA review and privacy-by-design checks |
| product-legal | Product counsel review for terms of service, privacy notices and feature launches |
| employment-legal | Offer letters, policy review and employment-agreement triage |
| ai-governance-legal | AI vendor review against an organization’s AI registry and model-risk checks |
| regulatory-legal | Regulatory feed watching, digesting and policy-diff tracking |
| ip-legal | IP clause review, licensing terms and infringement triage |
| litigation-legal | Claim charts, discovery support and case-management workflows |
| law-student | Study aids and doctrinal practice, not a commercial-team plugin |
| legal-clinic | Pro bono and clinic intake workflows |
| legal-builder-hub | Tooling for building new legal skills on top of the suite |
| cocounsel-legal (partner-built) | Thomson Reuters research and drafting tools; requires its own CoCounsel Legal subscription |
A commercial team installs four of these first:
A team already running a dedicated document analysis tool should weigh grounding and output format together rather than assume the plugin replaces it outright. The cocounsel-legal plugin sits apart from the other 11: it is vendor-maintained by Thomson Reuters, and it needs its own CoCounsel Legal subscription on top of a Claude plan.
Claude legal plugin commands are slash commands namespaced to their plugin, so contract review runs as `/commercial-legal:review` and a data subject access request response runs as `/privacy-legal:dsar-response`. The five commands below cover the recurring commercial work: review, triage, vendor checks, regulatory monitoring and data subject access requests. Each one reads the practice profile the cold-start interview wrote.
`/commercial-legal:review` runs a playbook-aware review of a vendor agreement, NDA or SaaS subscription. Run through the Claude for Microsoft 365 sidebar, it returns tracked changes in Word; run through Cowork or Claude Code, it returns a redline memo.
Sorting inbound NDAs still runs through `/commercial-legal:review`, which returns a green, yellow or red triage result. The plugin’s `/commercial-legal:escalation-flagger` command routes the ones that need attention to the right approver and drafts the ask.
Vendor contract checks combine `/commercial-legal:review` against the team’s own playbook with `/ai-governance-legal:vendor-ai-review` when the vendor is itself an AI tool, flagging playbook deviations and checking the vendor against the organization’s AI registry.
`/regulatory-legal:reg-feed-watcher` checks regulatory feeds and reports what changed since the last check, which functions as a regulatory digest rather than a raw feed. A companion scheduled agent, reg-change-monitor, runs the same check on a cadence with materiality filtering, rather than on demand.
`/privacy-legal:dsar-response` walks a data subject access request through to a drafted acknowledgment and substantive response within the statutory deadline.
The Claude legal plugin runs in three places:
Cowork is currently available on the Pro, Max and Team plans automatically, and Enterprise plans can opt in, per Anthropic’s own Cowork plan page. Anthropic is also folding the Cowork name into Claude, so a user on a Pro plan who now sees only “Claude” in the product is looking at the same surface this guide calls Cowork.
Claude Code installs the plugin from the marketplace: `/plugin marketplace add https://github.com/anthropics/claude-for-legal`, then `/plugin install commercial-legal@claude-for-legal` for the practice-area plugin a team needs. The same listing, claude.com/marketplace/plugins/legal, also shows what each practice-area plugin installs.
Anthropic authored the contract skills in the suite to work in the Claude for Word sidebar with tracked changes as the output mode, which is a change from the plugin’s January 2026 release, when it ran only in the Cowork desktop interface.
The Claude for Microsoft 365 sidebar puts the same skills inside Word, Excel, PowerPoint and Outlook from a single thread, so a lawyer can move from a Word redline to an Excel comparison without leaving the document. In Excel, a skill such as `corporate-legal:tabular-review` writes a multi-sheet workbook, with a sources sheet, rather than text in a reply. In Word, a reviewer accepts or rejects each change exactly as they would for a human markup, because the sidebar’s output is tracked changes rather than a chat reply.
The Claude legal plugin costs nothing to install or run on its own. The paid Claude plan it sits on top of starts at $17 per month on Claude Pro with annual billing, or $20 billed monthly, per Claude’s own pricing page. The plugins are open source and carry no license fee of their own, so the cost question is which plan and which connectors a team needs.
| Plan | Price | Notes |
|---|---|---|
| Free | $0 | not recommended for client-confidential work |
| Pro | $17/month (annual, $200 upfront) or $20/month (monthly) | |
| Max | from $100/month | |
| Team, standard seat | $20/seat/month (annual) or $25/seat/month (monthly) | |
| Team, premium seat | $100/seat/month (annual) or $125/seat/month (monthly) | |
| Enterprise | $20/seat/month (annual), 20-seat minimum | usage billed at API rates on top of the seat |
Two cost items the plan table does not show. First, connectors marked “customer subscription” in Anthropic’s own connector documentation need the customer’s own paid account and API key, billed separately from the Claude plan:
Second, Enterprise separates the seat price from consumption, so the monthly total moves with the models and tasks a matter runs.
For comparison, Spellbook prices per seat based on team size, with a free seven-day trial, and does not publish a standalone dollar figure.
To install the Claude legal plugin, download the Claude desktop app, open the Cowork tab, click Customize in the left sidebar, then click Browse plugins and install the practice-area plugins that match your work. Two steps after that decide whether the plugin produces anything useful: the cold-start interview, and a research connector. The four steps below cover the whole setup.
Download the app, open the Cowork tab, click Customize, then click Browse plugins. On Claude Code, install the same plugin with `/plugin marketplace add https://github.com/anthropics/claude-for-legal` followed by `/plugin install commercial-legal@claude-for-legal`.
Run `/<plugin>:cold-start-interview`, ten to twenty minutes per plugin, pointed at seed documents such as a signed master service agreement (MSA), a playbook or a prior review memo. Anthropic’s own documentation names skipping this step as a common reason a skill produces generic output. A quick-start option exists for a team that wants to try a skill before committing the full interview.
Connect CourtListener, Trellis, Descrybe or Solve Intelligence. Citations drawn from model knowledge alone are flagged `[verify]`, and with no research tool connected, the reviewer note on the final deliverable records that sources were not independently verified.
Edit each plugin’s `.mcp.json` for the contract lifecycle management (CLM) system, document management system, e-discovery platform or research source the team actually runs. Skills fall back to a narrower answer rather than silently skipping a step when a connector is absent.
The Claude legal plugin trades a configuration burden for openness: a team that runs the cold-start interview and pays for its own connector subscriptions gets a free, open-source suite it controls, while a team that skips the interview gets generic output. Neither path adds a market-wide benchmark to check a term against.
Pros
Cons
The Claude legal plugin and Spellbook now overlap on surface and differ on grounding: both return tracked changes in Microsoft Word, while Spellbook reasons from 20M+ real market contracts and enforces a firm playbook as a gate on every review. The Claude legal plugin arrives as a template a team configures; Spellbook arrives configured for commercial contract work.
| Axis | Claude legal plugin | Spellbook |
|---|---|---|
| Grounding | the model’s reasoning over the text, plus whatever the practice profile and connectors supply | 20M+ real market contracts, plus the team’s own playbooks and precedents |
| Market benchmarking | compares a term against the practice profile and any connected research tool | Compare to Market, which checks a term against the 2,300+ benchmark standards in the set |
| Playbook | a plain-English practice profile each skill reads, written by a cold-start interview per plugin | playbooks enforced on review, built by Customer Success in one to two weeks |
| Setup | ten to twenty minutes of interview per plugin, plus connector configuration | arrives configured, free seven-day trial, no implementation |
| Scope | any practice area, from law school to litigation | commercial contracts, not legal research or litigation |
| Cost shape | free plugins on a paid Claude plan, plus separately billed connector subscriptions | custom per-seat pricing by team size |
Both are genuine options for a commercial team. The suite being open source and free is a real advantage for a team willing to run the cold-start interview and own the connector setup. Before choosing, check three things:
Spellbook sits on the other side of that trade-off, pairing enforced playbooks with a contract set large enough to benchmark against, which is the choice a team weighing Claude for legal work is really making; it remains a contract specialist, not a legal-research or litigation platform.
Yes, the Claude legal plugin is free. Every plugin in Claude for Legal is open source and carries no license fee, and the cost of running one is the paid Claude plan behind it plus any connector subscriptions a team turns on.
Yes. Claude Cowork and Claude Code both run on Windows, macOS and Linux, and the Claude for Microsoft 365 sidebar runs anywhere Word, Excel, PowerPoint or Outlook does, including Windows.
A lawyer running the Claude legal plugin needs at least a Pro plan. Teams that need higher usage limits or centralized billing typically move to Max or a Team plan.
Attorneys can use the Claude legal plugin for client work if they treat every output as a draft and apply the same review and supervision obligations that already apply to any AI-assisted work product. The plugin does not change a lawyer’s underlying duty of competence or confidentiality.
Try Spellbook free to see how the Review feature handles redlines on your own contracts in Word, checked against your team’s own playbook.



.jpg)
Submission Received
Thank you for your interest!
Submission Received
Thank you for your interest!
We're connecting you with the best rep