CourtListener Scraper - Case Law & Legal Opinions API

Search the public CourtListener opinion corpus for US federal and state case law. Filter by full-text query, court, filing date, reporter citation, docket number, and publication status, then export source-linked JSON, CSV, or Excel records for legal research, monitoring, compliance, and RAG workflows.

The default run searches artificial intelligence copyright and returns at most 5 paid records. At the default 128 MB memory and live pricing, the maximum first-run charge is approximately $0.0101 ($0.0001 start + 5 × $0.002).

What each paid result contains

• case name, full case name, court, CourtListener court ID, and jurisdiction fields;
• filing and argument dates, docket number, docket ID, and opinion-cluster ID;
• judge/panel evidence, reporter/neutral/Lexis citations, and CourtListener citation count;
• publication status, suit nature, posture, procedural history, and syllabus when reported;
• every opinion document included in the search result, including type, author ID, per-curiam flag, cited-opinion IDs, snippets, and source/download links;
• exact source API URL, retrieval time, bounded query scope, candidate count, result rank, and mandatory coverage/interpretation notices;
• an explicit no_match evidence row when the bounded search returns no matching opinion.

Arrays are always emitted as [] rather than null, including on no-match output.

Input

Field	Description	Example
query	Full-text words or quoted phrases	artificial intelligence copyright
court	CourtListener court ID	scotus, ca9, dcd, nysd
dateFrom	Filed on/after, YYYY-MM-DD	2024-01-01
dateTo	Filed on/before, YYYY-MM-DD	2026-12-31
status	Exact Published or Unpublished result status	Published
citation	Exact reporter citation	410 U.S. 113
docketNumber	Exact docket number	70-18
orderBy	relevance, newest, or oldest	relevance
maxResults	Total paid-result cap, 1–5,000	5

status is enforced against the returned CourtListener status across paginated candidates rather than sent as an unsupported source parameter. maxResults is a total response cap, and the live buyer charge budget can lower it before the source is queried.

Examples

Recent Ninth Circuit opinions mentioning artificial intelligence

{
  "query": "artificial intelligence",
  "court": "ca9",
  "dateFrom": "2024-01-01",
  "orderBy": "newest",
  "maxResults": 10
}

Exact citation search

{
  "query": "",
  "citation": "410 U.S. 113",
  "maxResults": 5
}

Example output shape

{
  "setupStatus": "found",
  "caseName": "Electronic Privacy Information Center v. National Security Commission on Artificial Intelligence",
  "court": "District Court, District of Columbia",
  "courtId": "dcd",
  "dateFiled": "2020-06-01",
  "docketNumber": "Civil Action No. 2019-2906",
  "status": "Published",
  "citations": [],
  "opinionCount": 1,
  "opinionTypes": ["combined-opinion"],
  "opinionUrl": "https://www.courtlistener.com/opinion/4757780/electronic-privacy-information-center-v-national-security-commission-on/",
  "source": "CourtListener opinion search (Free Law Project public corpus)",
  "sourceRetrievedAt": "2026-07-25T00:00:00Z",
  "coverageNotice": "CourtListener is a non-governmental public archive assembled from many court sources...",
  "interpretationNotice": "This output is research evidence, not an official court copy, Shepardization/KeyCite treatment, legal advice..."
}

Fields absent from the source remain empty. The sample is shortened; live records use stable keys and include nested opinion-document evidence.

Pricing and bulk safety

Live pricing is:

• $0.0001 actor-start event per GB of reserved memory, minimum one event;
• $0.002 per default dataset item.

Results	Maximum item charge	Approximate total at default memory
5	$0.010	$0.0101
100	$0.200	$0.2001
1,000	$2.000	$2.0001

Output is persisted in size-bounded batches to avoid Apify's dataset request limit. Preview a narrow query before requesting a large corpus.

Evidence boundaries

CourtListener is maintained by the nonprofit Free Law Project; it is not a court and is not an official court-record system. Its corpus combines many sources, so coverage, metadata, citation counts, document links, and text can be delayed, incomplete, duplicated, corrected, or unavailable. A bounded no_match means only that this query found no match in the searched public corpus.

A CourtListener status or citation count does not establish that an opinion remains precedential, controlling, unmodified, or good law. The output is not an official court copy, Shepardization/KeyCite treatment, identity verification, legal advice, or a substitute for reviewing the linked opinion and current authoritative sources.

Source and reliability

• Source: CourtListener public REST API, maintained by the Free Law Project
• Authentication: none required for this public search workflow
• Reliability: retries request, body-read, rate-limit, and retryable HTTP failures with bounded backoff
• Scope: opinion/case-law search (type=o), not the separate RECAP federal docket archive