Register for a PACER account at pacer.uscourts.gov; PACER charges per-page fees for document access — budget accordingly and prefer CourtListener/RECAP for already-fetched public documents.
Use the CourtListener REST API (free, no charge per document) to search for cases via GET https://www.courtlistener.com/api/rest/v3/dockets/?court=&case_name= and retrieve docket entries via the docket entries endpoint.
For documents not yet in RECAP, use the PACER Case Locator (PCL) or court-specific CM/ECF query endpoints to search by case number, party name, or date range; authenticate with your PACER credentials via HTTP Basic Auth on each court's CM/ECF endpoint.
Download individual documents from CM/ECF using the document link obtained from docket entries; each PDF retrieval incurs PACER fees — deduplicate to avoid re-fetching already downloaded documents.
Contribute fetched documents back to RECAP via the CourtListener REST API (POST /api/rest/v3/recap/) to make them freely available and exempt from future PACER fees for other users.
Parse docket entry text using NLP to extract event types (hearing, order, motion filed), parties, and dates for downstream alerting or case tracking.
Known gotchas
PACER fees can accumulate rapidly in automated pipelines; set a per-run spending cap and track cumulative charges using the PACER billing API or manual account monitoring.
CM/ECF endpoints differ per court and are not fully standardized; selectors, form field names, and session management vary between district courts — scraping is fragile and may violate terms of service.
Some docket entries and documents are sealed or restricted; attempting to access them returns errors without indication that the document exists — never interpret a 404 or access error as confirming absence of a document.
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp