Recap MCP Server

RECAP — documents filed in US federal court cases

Direct links to the complaints, motions, orders and exhibits filed in federal cases. No API key, no PACER fee.

Part of Pipeworx — an MCP gateway connecting AI agents to 1866+ live data sources. This is an independent, unofficial integration — not affiliated with, endorsed by, or published by the upstream provider.

Auth

None.

How this is possible without a key

RECAP documents are not in CourtListener's bulk export. They are mirrored to the Internet Archive under an identifier that can be derived:

gov.uscourts.<court_id>.<pacer_case_id>

Federal docket coverage carries court_id for 100% of dockets and pacer_case_id for 84.8%, so a case's filings have a computable address. No agreement with anyone, no per-page charge.

What you get, and what you don't

You get: a link to each PDF, its docket-entry number and attachment number, its size, and the docket sheet where one exists.

You don't get text. Measured across 400 dockets: 141 archive items carry PDFs and none carry extracted text. Reading a filing means opening the document. Making these searchable would be an OCR project; this is not one.

Coverage is partial, and uneven

RECAP holds what PACER users have paid for and contributed, so coverage follows where lawyers use it. Measured over 400 dockets across 16 courts:

Overall: 35.8%.

The consequence matters more than the number: "no documents" means none have been contributed to RECAP — not that nothing was filed. Roughly two cases in three will come back empty, and every response says so. Treating an empty result as a fact about the court record would be wrong most of the time.

Tool

recap_filings — give it a docket_id from any docket search, or a court plus docket_number.

recap_filings({docket_id: 73433096})
  → V.O.S. Selections, Inc. v. Trump (cafc), 18 documents

Filenames encode the structure: ….12.0.pdf is docket entry 12, ….12.1.pdf its first attachment. That ordering is the only structure the archive provides — there are no document titles — so it is returned rather than hidden.

Related

• federal_civil_search, federal_criminal_search, federal_appeal_search, bankruptcy_search — find the docket first
• court-listener — opinions and full case text, a different corpus entirely

Data sources

• https://archive.org/ — RECAP archive items
• Federal docket coverage (CourtListener bulk data, Public Domain Mark)

RECAP exists because Free Law Project built it and PACER users contributed to it: https://free.law/donate/

Quick Start

Add to your MCP client (Claude Desktop, Cursor, Windsurf, etc.):

{
  "mcpServers": {
    "recap": {
      "url": "https://gateway.pipeworx.io/recap/mcp"
    }
  }
}

What this endpoint actually serves

tools/list at https://gateway.pipeworx.io/recap/mcp returns the tools in the table above plus the shared Pipeworx meta-tools — ask_pipeworx, discover_tools, search_within, remember/recall and the rest of the gateway-wide set. So the tool count you see is larger than this table: a single-pack endpoint currently lists roughly 30 shared tools alongside the pack's own. The connection's initialize response states its exact scope, and is the authoritative answer for a given day.

This is deliberate, not multiplexing by accident. The meta-tools are what let a scoped connection answer a question this pack does not cover — via ask_pipeworx, which routes across the whole catalog — without you adding a second MCP server. There is currently no way to mount a pack endpoint without them; if the extra schemas cost you more context than the routing is worth, connect to the full gateway once rather than to several pack endpoints.

Or connect to the full Pipeworx gateway to get every pack's tools listed directly, instead of just this one's:

{
  "mcpServers": {
    "pipeworx": {
      "url": "https://gateway.pipeworx.io/mcp"
    }
  }
}

Both URLs reach the same gateway and the same 1866+ data sources. The only difference is which pack's tools are listed directly; ask_pipeworx reaches all of them from either one.

No MCP client? Call it over HTTP

curl -X POST https://gateway.pipeworx.io/v1/tools/recap_filings \
  -H 'Content-Type: application/json' \
  -d '{"docket_id":73433096}'

No account needed for the first calls. Inspect any tool: GET https://gateway.pipeworx.io/v1/tools/recap_filings. Find one: POST https://gateway.pipeworx.io/v1/tools/search_packs with {"query":"..."}.

Run it from npm

@pipeworx/mcp-recap is a local stdio bridge to the Pipeworx gateway: your MCP client starts it with npx, and it forwards every call to the same https://gateway.pipeworx.io/recap/mcp endpoint shown above. It needs network access; there is no offline mode.

{
  "mcpServers": {
    "recap": {
      "command": "npx",
      "args": ["-y", "@pipeworx/mcp-recap"]
    }
  }
}

Without a key it calls anonymously, capped at 50 calls a day. A free account raises that to 200 a day: get a key at https://pipeworx.io/signup and add "env": { "PIPEWORX_API_KEY": "<your key>" } to the server entry. The key is sent as a Bearer token, the same way a hosted connection sends it, so calls are metered and billed exactly as they are there.

It lists only this pack's tools by default. Set PIPEWORX_META_TOOLS=1 to also list ask_pipeworx and the other gateway-wide tools.

Behind an HTTPS proxy: Node's built-in fetch does not read HTTPS_PROXY by default. Newer Node releases honour it when NODE_USE_ENV_PROXY=1 is set; check your Node version's documentation.

Using with ask_pipeworx

Instead of calling tools directly, you can ask questions in plain English — this works on the pack endpoint above as well as on the full gateway:

ask_pipeworx({ question: "your question about Recap data" })

The gateway picks the right tool and fills the arguments automatically.

More

• Docs and guides
• pipeworx.io

License

MIT