PACER Court Records Intel - MCP Server

Model Context Protocol (MCP) server wrapper for court-records-intel. Gives AI agents direct, pay-per-call access to federal litigation intelligence - PACER-derived dockets via the free RECAP archive, with no PACER login and no PACER per-page fees.

What it is

A thin MCP server that exposes the federal-litigation engine as five typed tools an AI agent can call. Each tool runs the upstream court-records-intel actor (CourtListener API + RECAP archive) and reshapes the result into small, deterministic JSON. Built for legal-research copilots, litigation-finance agents, M&A due-diligence bots, and competitive-intel workflows.

Tools

Tool	What it does	Key args
get_party_litigation	Federal cases for a person or company (plaintiff or defendant) + a rollup (total/active/closed, top courts, settlement-ratio estimate).	party, courts[], date_from, limit
get_litigation_portfolio	Corporate-entity aggregate: jurisdiction mix, top judges, top opposing counsel, exposure estimate, diligence red flags. Rollup record returned first.	entity, date_from
get_attorney_track_record	An attorney's case load by year, top practice areas, opposing counsel, judges appeared before, win/loss estimate.	attorney
get_recent_filings	New federal dockets filed since a date, optionally scoped to courts / a party watchlist / keywords. The new-case alert tool.	since, courts[], parties[], keywords[], limit
get_case_documents	RECAP document URLs + metadata for a case number; include_text=true also returns extracted filing text.	case_number, include_text

Every tool returns the same envelope: { tool, args, ok, count, items, summary, error }. Agents key off summary for natural-language framing and items for structured downstream use. Every tool also accepts an optional courtlistener_api_key (a free key raises CourtListener rate limits from ~60 req/min to ~5000/hr).

Run modes

• list_tools - emit the tool catalog (free, no charge) including the agentic-payment descriptor.
• call_tool - invoke one tool. Requires tool + args.
• batch - invoke up to 10 {tool, args} calls in one run.

{
  "mode": "call_tool",
  "tool": "get_litigation_portfolio",
  "args": { "entity": "Helion Compute, Inc." }
}

You can also set courtlistener_api_key at the top level of the input and it flows into every tool call (unless a per-call args value overrides it).

Agentic payments (x402 + Skyfire)

This server is x402 (USDC on Base) and Skyfire ready. When the operator enables Apify MCP monetization, an AI agent can pay per tool call with no pre-provisioned API token. The list_tools response includes a payments descriptor advertising the accepted rails and per-call price. Operators enable rails via environment variables:

Env	Purpose
X402_ENABLED	1 to advertise x402 acceptance
X402_PAY_TO_ADDRESS	receiving wallet (USDC on Base)
X402_PRICE_USD	per-call price advertised (default 0.005)
SKYFIRE_ENABLED	1 to advertise Skyfire acceptance
SKYFIRE_SELLER_ID	Skyfire seller identity

Calls through Apify always bill via standard PPE (mcp_tool_call $0.005 + upstream pass-through); x402/Skyfire are the token-less rails layered on top for direct agents.

Pricing

Flat $0.005 per MCP tool call, plus the upstream court-records-intel PPE pass-through billed to the same run:

• case_record: $0.005 per docket (parties, attorneys, status, judge, recent entries)
• document_record: $0.003 per RECAP document URL (when documents are requested)
• document_text_charge: $0.010 per document with extracted text (include_text=true)
• intelligence_record: $0.015 per aggregation (portfolio or attorney rollup)

PACER charges $0.10 per page for the same filings; RECAP already liberated them, so there are no PACER fees here. mode=list_tools is free, and a tool call that returns nothing costs only the $0.005 wrapper fee.

Coverage honesty

RECAP mirrors only what PACER users have already pulled and contributed, so coverage is strong for high-profile and frequently-accessed dockets and thinner for obscure ones. When a docket or document is not in RECAP, the upstream actor flags the gap with available=false and a human-readable reason rather than guessing.

Responsible use / data scope

This server reads only logged-out, public federal court records via the CourtListener API and the RECAP archive. No PACER login, no accounts, no cookies, no paywall bypass. Court dockets are public records; you are responsible for lawful use of the outputs (for example, FCRA limits apply if you use these records for employment, credit, or tenant screening). See the upstream actor's README for the full data-scope note.

Related Actors

• court-records-intel - the non-MCP actor with full input controls, monitor mode for watchlist alerting, and CSV/HTML artifacts.
• mcp-sec-edgar-intel - pair corporate litigation exposure with SEC filing intelligence in the same diligence agent.

Repo

Source: portfolio repo, actors/mcp-court-records-intel/. Upstream: actors/court-records-intel/.

Found this useful?

Leave a quick review: https://apify.com/seibs.co/mcp-court-records-intel#reviews