Case Harvester is back
Case Harvester started in 2018 as an Open Justice Baltimore project. Maryland publishes court records through Case Search, but the site is designed to retrieve one case at a time. It does not let a researcher study the court record as a whole, and therefore prevents research of systemic patterns. Case Harvester collects those cases into one database so researchers, legal-aid organizations, and journalists can do that work.
The first iteration of Case Harvester was successful in achieving these goals, especially for researchers studying pretrial detention, bail, and eviction. The Legal Data & Design Clinic at the University of Baltimore School of Law and faculty at Maryland Carey Law used the data for “A Butterfly in COVID”, a Maryland Law Review study of Baltimore’s pretrial system. The University of Maryland’s Crime Research and Innovation Center used it in a statewide pretrial risk assessment. The Princeton Eviction Lab used it in its Eviction Tracking System, the Howard Center for Investigative Journalism used it to follow evictions after the pandemic moratorium ended, and Court Watch PG used it to track eviction filings and bail hearings. The ACLU of Maryland, the Lawyers’ Committee for Civil Rights Under Law, Trans Maryland, and The Appeal used our database, as did students at Maryland and MIT.
Not long after Case Harvester began publishing these records, the Maryland Judiciary started implementing measures to prevent scraping and other automated access to Case Search. While these measures were largely trivial to bypass due to poorly designed and broken implementations, eventually the maintenance burden and lack of funding for Case Harvester led to the decision to put the project on indefinite hiatus around May 2023.
Today we are excited to announce that the hiatus is finally over, and we are relaunching a brand new version of Case Harvester here at caseharvester.org.
What changed at the courts
When Case Harvester began, Maryland’s courts were partway through a ten-year move to a single statewide case management system, Maryland Electronic Courts, better known as MDEC. The rollout went county by county starting in 2014. Baltimore City, the largest and last jurisdiction, joined in May 2024.
Until that finished, Case Search served each court’s cases in its own layout. There were eighteen of them, each with a code the Judiciary never published. We learned the list through a public records request. Baltimore City criminal cases looked nothing like Montgomery County criminal cases, which looked nothing like a District Court traffic citation. Every layout needed its own parser, and a change to any one of them could break the continual collection of new case data.
In 2026 the Judiciary replaced Case Search itself. The new Case Search and Record Portal is built on the Odyssey platform from Tyler Technologies, the vendor behind MDEC. Underneath the new pages, the portal serves each case as structured data rather than as a web page to be tediously parsed. This brought Case Search into alignment with how most websites on the internet have been operating for nearly two decades, and so was a welcome improvement. It also allowed us to capture and represent case data far easier and with much higher fidelity, and is what led us to decide to rebuild Case Harvester from the ground up rather than iterate on the old design.
What changed in Case Harvester
Every case in Case Harvester now has one consistent structure, whatever court it came from or when it was filed. We converted the archive collected since 2018 into that same structure, so a Baltimore City case from 2003 and a case filed last week can be studied together.
Coverage reaches back to 2000 statewide and to 1990 for Baltimore City. New filings come from the Judiciary’s daily new-case reports. The Judgment and Liens Search adds more case numbers once a judgment appears. We re-check existing cases on a schedule that follows their age and activity. The statistics page shows the live numbers, coverage by category and year, and how current the record is.
The database now holds more than cases. Case Harvester collects the Judiciary’s five official ID lists of judicial officers, its attorney directory with the courts’ published sanctions from fiscal year 2006 on, and the reported opinions of both appellate courts from 1995 on. Each connects back to the cases. A judge’s record links to the cases they appear on, an attorney’s listing to their appearances in cases, and an appeal to the opinion that decided it.
Here is what that means in practice. Suppose you want every case a particular judge has sat on in the past five years. On Case Search you would collect those one case at a time. In Case Harvester, open the judge’s record and the cases are listed under it.
Some things are redacted on purpose. By default, the data we serve omits the names, exact dates of birth, street addresses, and physical descriptions of criminal defendants and eviction tenants. Cases and charges we collected before they disappeared from Case Search remain available to customers and are marked. And now that AI has made bulk data collection cheaper and more profitable for both mega-corporations and bad actors, we decided to only allow access to our database by vetted organizations and individuals. See our Responsible Use section of the website and the data privacy page for more details.
What a license includes
Every license includes three ways to access the data:
The web app. Browse every published table in your browser, filter and sort it, click on a case number to view the full structured record, and export filtered case records as a spreadsheet.
The API. Your own code can access the same records, and we provide a written guide and interactive reference for how to do that. Filter counts, grouped totals, case lookup, and exports work the same way they do in the app.
The MCP server. Point your own AI agent like Claude Code or ChatGPT Codex at Case Harvester and it can look up cases, filter tables, and count matches directly. The connection guide walks you through setup. The same terms apply.
Access to the new site is licensed, and we personally read every application. Funded research pays a flat price that does not depend on how much data you access. Independent researchers, students, community advocates, small legal-aid organizations, and journalists working without dedicated funding can pay what they can afford. The access section has the details.
Access can be requested at caseharvester.org/request-access. Questions go to [email protected].