A (free!) Legal Connector that Gets Out of the AI’s Way.
DingDuff is a Claude connector for legal research. It links your existing Claude account to a database of millions of legal opinions, state and federal statutes, and court filings from around the country.
We’ve designed DingDuff to be simple, useful, and inexpensive to run, which allows us to currently operate it with a tip jar. All our effort has gone into one thing: building a tool we actually use in our own daily work. That meant making it run seamlessly inside Claude. We don’t dress it up in fancy packaging or try to tell the AI how to do its job. Our goal is to empower Claude to do what we ask, then get out of its way — and we’ve found the results to be quite brilliant when we do.
Legal Sources in DingDuff
More About DingDuff
DingDuff started as a side project by two practicing lawyers — us, Kyle and Stephanie — who wanted something we could use at work in our own day jobs. We were never trying to build a company.
What we didn’t expect was to get an answer to an empirical question that we hadn’t even thought to ask: does a purpose-built legal AI platform (like Westlaw’s AI/Deep Research) actually perform better than a frontier AI model (like Claude Opus) that has been given direct access to the legal sources?
In our hands, DingDuff has outperformed our firms’ Westlaw AI accounts so consistently that we both quietly stopped using them. That’s when we decided to make DingDuff available to other lawyers.
Who Is DingDuff For?
DingDuff is for lawyers. The reason DingDuff works so well is because we haven’t tried to program a bunch of guardrails into it, or teach Claude how to lawyer. But that same hands-off design means Claude + DingDuff will occasionally get something wrong (although we have never seen it hallucinate a fake case or statute).
With DingDuff, the lawyer reading and checking the output acts as the guardrail. That’s why having a lawyer-user is essential, and the reason we’re making DingDuff available only to lawyers.
By signing up for and using DingDuff you accept the Terms of Service, which includes a promise that you’re a lawyer.
Get DingDuff
You need two things to get started: (1) your own Claude plan and (2) a username/password for the DingDuff “connector.”
1 Sign Up for DingDuff
2 Install DingDuff
Open Claude, paste this link into it, and ask Claude to walk you through the installation: github.com/DingDuff/dingduff-plugins/wiki
Or, if you’d prefer to install it manually, you can visit the wiki directly and follow those instructions. Or you can review our manual installation page here.
How DingDuff works
DingDuff works wherever Claude does — and it adapts to where you’re running it.
DingDuff can run right inside Claude’s web app and mobile app. Because Claude can’t save files to your device here, you’ll have to make sure the virtual machine feature is turned on in your Claude settings.
To do this, go to your Claude account → Settings → Capabilities. Scroll down and turn on two features: “Claude code execution and file creation” and “Allow network egress.”
DingDuff is at its best in Claude Cowork, where Claude can reach a folder on your computer.
Instead of pulling opinions into the chat, it saves your own full copy of each opinion to a desktop folder so that you can revisit them later.
DingDuff works on its own, but our free skills make its answers noticeably better. They’re one-time downloads that stay in your skill picker once uploaded. Download them directly here, or grab the latest versions from our install wiki.
Systematic case-law research for everyday questions — finds and analyzes relevant opinions, verifies holdings, quotes, and citations, and surfaces splits. The right choice for most work.
Exhaustive, recursive research that maps the full citation network — back to the doctrinal source and forward to the latest applications — with a validity check on every key case. For briefs and formal opinions.
Exhaustive statutory research — maps a code top-down and bottom-up, discovers definitions and cross-references, and pulls judicial interpretations of ambiguous or controlling language.
Verifies every citation in a drafted memo (Markdown, Word, or Google Docs) against stored opinions and statutes, then opens an interactive attorney review panel and records your verdicts.
Formats citations the way courts expect — signals, pincites, short forms (id./supra), string cites, and same-matter filings.
By default, Claude uses the standard research skills. The deep skills use a lot more tokens, so you have to ask Claude to use them explicitly — it won’t run them on its own. More is explained here.
Costs — Tip Jar
DingDuff is free. But we do have some costs in running it. Thankfully our costs are low enough that we can be supported through a tip jar. If you find DingDuff helpful and would like to help keep it free, please consider donating.
Processed securely by Stripe. No account required.
Meet the Team
Like most lawyers these days, we (Kyle and Stephanie) felt acutely the twin fears that (a) the robots would eventually replace us, and (b) in the meantime, the other lawyers would get a better robot and use it against us. So before Westlaw AI even existed, and anticipating that other lawyers would soon have some sort of AI death star we'd have to confront, we began developing our own tool for our own use.
For a year, we've made iterative improvements. DingDuff began as a scrappy interface to a database we had compiled locally, running on a Rasberry Pi tucked away in Kyle's closet. But it has since grown into something far more capable—a powerful legal tool that runs through several databases, including CourtListener** and state and federal statutes. And after two years of dreading the inevitable Westlaw death star, we discovered a happy accident: our homegrown tool works comparably well, and often better in our opinion.
Finding ourselves unexpectedly in possession of an incredibly useful legal tool with significant potential, we wanted to share it with our lawyer friends. To help turn DingDuff from a personal project into a shareable product, we brought on two software developers: Jeremy, a longtime friend from our undergraduate philosophy days at UT Austin, and Julia, a recent computer science graduate.
**This entire project would never have been possible without the data heroes at CourtListener.
Kyle is a mass torts lawyer with a passion for history, philosophy, the law, and everything Reddit. He has served as first or second chair in more than two dozen trials and written and argued appellate briefs before the Texas Supreme Court and federal courts of appeals. But in his legal career, he's derived most of his joy from representing underserved tenants in landlord-tenant disputes. He has an insatiable curiosity and a gift for solving complex logic puzzles — which together have been the driving forces behind DingDuff. Kyle grew up in the Austin area, where his dad was a software engineer and his mom a lawyer. He's a proud double longhorn (both undergrad and law school) and the main brain behind DingDuff.
Stephanie is a commercial litigator with a background in financial regulation and litigation. She grew up in the Austin area in a family of engineers, scientists, and one very skilled carpenter. Since law school, she has practiced in private firms, served in the federal government, clerked for a federal judge, and taught legal research and writing at Georgetown Law. She has her law degree from Columbia Law School.
Julia is a software developer who is always finding ways to pair her computer science degree from Texas A&M with her passion for art and design. Julia also grew up in Austin and comes from a family of computer scientists.
George is Stephanie's dog and the DingDuff fetcher of legal sources.
Contact Us
Have a question, feedback, or just want to say hello? We'd love to hear from you.
Email Us