We Mapped Five Legal Data APIs by What They Actually Know. Only One Knows Your Margins.
"Legal data API" is doing a lot of work as a category name. PACER, RECAP, UniCourt, and Trellis all get filed under it, and they answer almost none of the same questions.
Record breadth vs. judge-specific procedural depth
Qualitative placement based on each vendor's own public product description, not a scored benchmark
"Records" means dockets, opinions, and case data. "Rules" means structured, citable, judge-specific filing requirements, page limits, and formatting constraints, which is a different thing than analytics about how a judge tends to rule.
Every product on this list is good at the thing it was built for, which would be more reassuring if they had all been built for the same thing. Here's what each one actually returns when you query it, so you can tell which one you need before you sign anything.
PACER: the record of record, priced like it's 1998
PACER is where federal docket entries and filed documents legally live. It's the most complete federal record there is, and it charges $0.10 a page for the privilege of reading it, capped at $3.00 per document, a fee schedule that hasn't moved since 2020. There's no modern token-based REST API for pulling docket content, no webhooks, no delta feed. PACER's change-detection system is you, logging back in, ten cents a page. It knows everything that happened in a case and nothing about what a judge requires before you file the next thing.
CourtListener: the free archive with the biggest heart
Free Law Project's CourtListener is a donation-funded 501(c)(3) platform that liberates PACER data through its RECAP extension and layers on case law, judge biographical data, and financial disclosures, all queryable through a REST API that's free with a membership. For knowing who a party is, what a judge has published, or when a document was filed, it is excellent, and it pays for all of it by asking nicely. It has no field for a page limit, because a docket was never designed to hold one.
UniCourt: 4,000+ courts, price on request
UniCourt is the biggest number in this post: 4,000+ state and federal courts across 40+ states, by its own count, with a normalized dataset it markets as the largest structured litigation database in the country. Its self-described API-first platform and newer DEEP product are built for business development and litigation strategy: docket automation, party and firm analytics. The buyer is a BD team, not an associate double-checking a standing order. Pricing is enterprise and not published. It's a data-breadth product, not a judge-compliance product, and to its credit, it doesn't claim to be.
Trellis: knowing how a judge rules, not what they require
Trellis focuses on state trial courts and sells judge analytics: how a specific judge tends to rule on a given motion, verdict history, an opposing attorney's track record. Entry pricing starts around $69.95 a month, with a Lawyer plan around $120. That buys a different meaning of the phrase "know your judge." One tells you how a judge is likely to decide. The other tells you whether your filing gets read at all.
The side-by-side
| Product | What it actually is | Judge filing rules? | Pricing |
|---|---|---|---|
| PACER | Federal docket records, filed by the courts themselves | No | $0.10/page, $3.00 cap per document |
| CourtListener | Free case law, docket data, and judge bios (Free Law Project, nonprofit) | No | Free with membership; commercial agreements available |
| UniCourt | Normalized docket data and litigation analytics at scale | No | Enterprise, not published |
| Trellis | State trial court judge and litigation analytics | No (analytics on outcomes, not filing format) | From ~$69.95/mo |
| Court Rules | Judge-specific filing compliance, cited to page and section | Yes | Plan-based, REST + MCP |
The last question before you hit file
None of the four products above is a bad answer. They're answering "what happened" and "what's this judge like." We built Court Rules to answer the question that comes right before you hit file: does this specific document, for this specific judge, in this specific court, meet the rule that's about to get it rejected. Our coverage number is the smallest in this post: 630+ judges across 20 districts. Small on purpose, because the only coverage number that matters is whether the judge reading your filing is on the list.