Court Records Scraper for CourtListener

Search US court records from CourtListener and export federal dockets, case law opinions and RECAP court documents as clean rows. Built for legal tech teams, due diligence and KYC analysts, litigation researchers and paralegals who need a court docket lookup by party, court, docket number or filing date, with optional party roles and attorney contacts, without writing code against an API.

What this court records scraper does

CourtListener is a public legal research service run by the non-profit Free Law Project. It holds millions of US court opinions and the RECAP archive of federal court filings that people have retrieved from PACER. This Actor turns a CourtListener search into a dataset you can download as JSON, CSV or Excel, or pull through the Apify API.

You choose what to search:

• Dockets: federal case files with case name, docket number, court, assigned judge, filing and termination dates, cause, nature of suit, parties and attorneys.
• Opinions: case law from federal and state courts with citations, citation count, argued date and publication status.
• RECAP documents: individual filings and attachments with document number, description, page count and a PDF link when one is available.

Optional details add each party's role (Plaintiff, Defendant, Debtor and so on) and the attorneys who represent it, with firm, address, phone and fax as listed on the public record.

Which court records are covered

The Actor searches what CourtListener has indexed. Federal RECAP dockets depend on documents contributed to RECAP and can be incomplete or older than the latest PACER activity. State case law is included where CourtListener has it, but this is not a search of every state or county trial court. Unavailable values stay empty; the Actor never infers case outcomes, party identities or legal conclusions.

Input

Provide at least one of query, party_name, case_name, docket_number or court. Every other field narrows the search.

Field	Type	Default	What it does
query	string		Keyword query with CourtListener syntax: quoted phrases, AND/OR, field:value
search_type	string	dockets	dockets, opinions or recap_documents
court	list of strings		CourtListener court IDs, for example ["txed"] or ["txed", "cand"]; API callers may send one ID as a string
filed_after / filed_before	date		Inclusive filing date range, YYYY-MM-DD: document date for RECAP documents, case date for dockets, opinion date for opinions
party_name	string		Party name filter (dockets and RECAP documents)
case_name	string		Case name filter
docket_number	string		Docket number; combine with court, because the same number exists in several courts
nature_of_suit	string		Nature of suit code or text, for example 830 for patent (dockets and RECAP documents)
judge	string		Judge filter (opinions)
opinion_status	string	source default: Published	Published, Unpublished, Errata, Separate, In-chambers, Relating-to, Unknown
available_only	boolean	false	RECAP documents: only filings available in RECAP
sort	string	relevance	relevance, date_desc (newest filing first) or date_asc, using the same date as the filing range
max_items	integer	50	Maximum primary results, up to 1,000; optional document rows come on top
include_details	boolean	false	Dockets: party roles and associated attorney contacts
include_documents	boolean	false	Append nested filing matches (dockets) or opinion documents (opinions) after each case
cursor	string		next_cursor from an earlier run with identical input, to continue a large search

include_documents returns links and metadata; it does not download PDF files. Example input, the newest patent cases in the Western District of Texas:

{"query":"patent","court":["txwd"],"nature_of_suit":"830","filed_after":"2026-01-01","sort":"date_desc","max_items":2}

Output

The dataset has two row types, marked in record_type (and _type): court_case and court_document. Every row carries row_id, source and a UTC fetched_at timestamp for the request that retrieved it; continued pages receive a fresh timestamp. row_id is stable and unique across dockets, opinion clusters, RECAP documents and opinion documents, so you can deduplicate continued searches safely. Document rows follow their case when you request documents.

This is a real row returned for the example input above, captured from CourtListener on 2026-10-05:

{
  "record_type": "court_case",
  "row_id": "case:docket:74910277",
  "kind": "docket",
  "case_id": "74910277",
  "case_name": "Flash Uplink LLC v. Rapid7, Inc.",
  "docket_number": "7:26-cv-00401",
  "court": "District Court, W.D. Texas",
  "court_id": "txwd",
  "judge": "Walter David Counts III",
  "filing_date": "2026-10-02",
  "termination_date": null,
  "cause": "35:271 Patent Infringement",
  "nature_of_suit": "830 Patent",
  "jurisdiction_type": "Federal Question",
  "jury_demand": "Plaintiff",
  "parties": [
    {"name": "Flash Uplink LLC", "role": null, "attorneys": []},
    {"name": "Rapid7, Inc.", "role": null, "attorneys": []}
  ],
  "attorneys": ["Bryan N. Dematteo"],
  "firms": ["DeMatteo Law LLC"],
  "citations": [],
  "details_fetched": false,
  "url": "https://www.courtlistener.com/docket/74910277/flash-uplink-llc-v-rapid7-inc/"
}

Case rows contain case_name, case_name_full, case_id, docket_id, cluster_id, docket_number, court, court_id, court_citation, judge, panel, filing_date, termination_date, argued_date, cause, nature_of_suit, referred_to, jurisdiction_type, jury_demand, chapter, trustee, pacer_case_id, parties, attorneys, firms, citations, opinion_status, cite_count, more_documents, details_fetched and url. An opinion row, for example Cqv Co., Ltd. v. Merck Patent Gmbh in the Federal Circuit, carries the citation 130 F.4th 1344, its argued date and status Published.

Party names come with every docket; their role stays null until details are retrieved. With include_details, details_fetched becomes true and each party lists its attorneys with name, firm, address, phone, fax and designations (for example ATTORNEY TO BE NOTICED) where the public record shows them. Email addresses are not promised. Opinion rows keep attorneys empty because the source provides unstructured text with firms and roles. That original text is preserved as attorney_text (string or null); docket rows have attorney_text: null.

Document rows contain document_id, case_id, docket_id, case_name, docket_number, court_id, document_type, document_number, description, long_description, filing_date, page_count, pdf_url, url, docket_entry_id, entry_number, attachment_number, is_available, pacer_doc_id, snippet, per_curiam and source_download_url. A snippet is an excerpt, not the full text.

The run output (key-value store record OUTPUT) holds row counts, next_cursor and warnings. A failed detail lookup keeps the case row and names the docket in a warning. An unusable source record is skipped with a warning and does not count towards max_items or charges; other cases and documents are retained. Missing values are null and missing lists are empty arrays, so absence is never confused with an invented value.

Use cases

• Due diligence and KYC: check whether a company or person appears as a party in federal litigation or bankruptcy before a deal, loan or onboarding.
• Litigation monitoring: schedule a daily run with sort: date_desc on a court and nature of suit (for example patent cases in E.D. Texas) and get new filings as rows.
• Legal tech and AI: feed case law, citations and docket metadata into search, RAG or analytics products through the Apify API.
• Paralegal research: export a docket lookup by docket number and court, with document descriptions and PDF links, into a spreadsheet.
• Business development for law firms: find recent cases by nature of suit and see which firms and attorneys appear on them.

How it compares

Option	What you get	Trade-off
PACER	The official federal record, newest filings	Paid per page, one court at a time, no bulk export
CourtListener website	Free search of the same collections	Manual, page by page
CourtListener API directly	Full flexibility	You write and maintain the client: authentication for some endpoints, paging, rate limits, party page parsing
Other court records actors	Often one collection (only opinions or only dockets)	Fewer filters, no party roles or attorney contacts, or a higher price per row
This Actor	Dockets, opinions and RECAP documents in one schema, party roles and attorney contacts, cursor continuation	Coverage is CourtListener's; it does not buy unavailable PACER documents

Pricing

You pay per delivered row, at the rates shown in the Actor's Pricing tab:

• court_case: each delivered case row (docket or opinion cluster).
• court_document: each delivered document row, so include_documents adds document charges.
• case_details: each delivered case whose public party details were retrieved successfully. A failed detail lookup is not charged.

There is no start fee, and a search that returns no rows costs nothing. Free users receive at most ten dataset rows per run, document rows included, and are charged only for the rows actually delivered. Paying users collect up to max_items primary results plus any selected document rows, within their run spending limit.

FAQ

Does this search every US court?

It searches CourtListener's collections: federal RECAP case files and indexed federal and state case law. Local and county court coverage varies. For a complete case file, compare with the court itself or PACER.

Can I get every document from a docket?

include_documents adds up to three matching filings per docket, not the full docket. Use search_type: recap_documents with query: "docket_id:45814286" to search all filings of one case. A PDF link is returned only when the source has one; the Actor does not purchase PACER documents.

Are full opinion texts included?

No. Opinion metadata, citations, snippets and available PDF links are included. Download the linked PDF if you need the full text.

Why are party roles or attorneys missing?

Roles and attorneys come from the public parties page, which you request with include_details (dockets only). Some cases list no parties or representation, and a lookup can fail. Check details_fetched and the run warnings; the case row stays, and you are not charged for details that were not delivered.

How do I continue a large search?

The Actor follows continuation pages automatically up to max_items. In API integrations you can pass next_cursor from the run output as cursor with the same input. Stop when next_cursor is null.

Do I need a CourtListener account or API token?

No. You only need an Apify account; the Actor handles the requests and returns the rows.

Source

Free Law Project's search API documentation describes the collections, publication statuses and query syntax. Keep the CourtListener url with exported data so each case or document can be checked against its source.