Last week, Anthropic released a legal plugin for Claude Code and Claude Cowork.

It's open-source, markdown-based, and requires no installation, no database, and no separate account.

You get six features out of the box: contract review, NDA triage, vendor checks, legal briefings, template responses, and background compliance knowledge.

It connects to your existing tools (Slack, Box, SharePoint, Atlassian) via MCP integrations.

One thing worth noting: you do need a Claude Pro or Max subscription to access it.

The plugin itself adds no cost on top of that, but it's not quite "free" in the absolute sense.

As you've probably seen with most AI tools, the real question isn't "can it do something" but "can it do something useful."

This one can.

Contract review runs clause-by-clause analysis with GREEN, YELLOW, and RED flags across liability caps, indemnification, IP, data protection, confidentiality, termination, and more.

It tells you where the contract deviates from standard positions and gives you specific redline language to propose back.

NDA triage checks 13 screening criteria and routes to GREEN (sign it), YELLOW (quick counsel pass), or RED (needs full review).

Vendor check pulls every agreement you have with a given vendor, shows what's in place versus what's missing, and flags anything expiring within 90 days.

For a tool that requires zero implementation, this is genuinely impressive.

I've recorded a full walkthrough showing you how to use every feature: contract review, NDA triage, vendor checks, the lot.

See exactly what it does and where it falls short, so you can decide whether it fits your workflow.

Here's where the honest part comes in.

Every conversation starts from scratch. No memory of your last review, your previous positions, or how your team handled a similar clause six months ago.

There's no audit trail, so you can't prove what was reviewed or when.

No export to Word or tracked changes. Everything lives in the chat window.

No human-in-the-loop approval workflow or escalation chain.

No playbook versioning. Your positions live in a single markdown file with no rule hierarchies.

No clause-level extraction, no contract comparison between versions, and no multi-user support.

These aren't edge cases.

These are the features that separate "useful assistant" from "production-grade legal AI."

The 80% handles the routine.

The 20% handles what happens when something goes wrong.

No audit trail means no defensibility.

No memory means no institutional knowledge.

No approval chain means no governance.

If you're managing compliance across an organization, these aren't nice-to-haves.

If you're thinking about risk exposure across your firm's matters, these are table stakes.

An estimated 80% of legal teams exploring AI agents still want human-in-the-loop oversight, according to LegalOn's January 2026 survey.

That instinct is correct.

The 20% is where legal risk, regulatory exposure, and competitive advantage actually live.

Two routes to the last 20%.

Dedicated legal AI platforms like Harvey, LegalOn, and others run several thousand dollars per seat per month, built for production-grade work with audit trails, compliance features, and structured workflows.

The other route is custom AI development. Your playbooks, your integrations, your audit trails, your approval workflows, built around how your team actually operates.

We recently built a contract review platform for a client that does exactly this: clause-level extraction with confidence scoring, playbook rules with severity hierarchies, tracked-changes Word export, human accept/reject decisions on every suggestion, and a full audit trail.

That's the kind of gap a custom build fills, and it's the kind of work we do at Cyberaktive.

The right path depends on your team size, risk profile, and how much of the workflow you need to control.

Apply the 80/20 lens to any legal AI tool you evaluate.

The 80%: Can it handle first-pass review, triage, and research?

Most tools can, and this free plugin proves it.

The 20%: Does it have audit trails, governance, export, memory, and human oversight?

That's what separates useful from production-grade.

Use this plugin to see what's possible.

Then decide what your risk profile actually demands.

How is your team currently handling AI for contract review?

A) We use a dedicated legal AI platform

B) We use general-purpose AI (ChatGPT, Copilot, etc.)

C) We're evaluating tools but haven't committed

D) We don't use AI for contract work yet

[Reply with your letter choice] - I'll share the results in the next edition.

Look, this plugin is a milestone.

Free, capable, and genuinely useful for first-pass legal work.

But it's a starting point, not an endpoint.

The firms and legal teams that get the most from AI are the ones who understand where the free version stops and what the next level requires.

Try it.

See what it does.

Then make a deliberate decision about the 20%.

Hit reply and tell me: have you tried this plugin yet?

What did you think?

— Liam Barnes

We help legal teams figure out where off-the-shelf AI stops and custom solutions start. AI audits, workflow automation, and custom builds.

Grab some time to chat

(if you don’t see a suitable time, just shoot me an email [email protected])

How Did We Do?

Your feedback shapes what comes next.
Let us know if this edition hit the mark or missed.

Too vague? Too detailed? Too long? Too Short? Too pink?

👍 Loved it | 👎 Not for me | 💡 I Have a Suggestion

OR: [email protected]

Was this week’s newsletter forwarded to you?

Sign up, it’s free.