Hudoc MCP server

@pipeworx/hudoc

European Court of Human Rights (ECHR / Strasbourg) case law — 231,000+ judgments, decisions, advisory opinions, resolutions and communicated cases across the 46 Council of Europe states, served live from the Court's own HUDOC database.

Part of Pipeworx — an MCP gateway connecting AI agents to 1576+ live data sources.

Tools

• echrsearch(query?, article?, respondent?, collection?, importance?, datefrom?, date_to?, limit?) — search case law by free text, Convention Article, respondent state and/or date. Returns case name, application number(s), Article(s), conclusion (violation/no violation) and the itemid to read the full text with.
• echrjudgment(itemid, maxchars?, offset?) — full text of one document by its HUDOC itemid, paged.

Auth

Keyless. No registration, no API key.

Data sources

• — the search API HUDOC's own web UI calls. Query syntax is a Lucene-ish field grammar (contentsitename=ECHR AND (documentcollectionid2:JUDGMENTS) AND article=8 AND respondent=FRA AND kpdate>=2024-01-01 AND "surveillance"). Free text is a bare double-quoted phrase with no field name — a field-qualified phrase like conclusion="freedom of expression" silently returns zero results.
• — the document body, converted from the Court's stored DOCX to HTML. This is NOT the /eng?i=... page URL: that URL is a static Angular SPA shell — same ~15.7KB byte-for-byte regardless of itemid, including nonexistent ones — and never contains the judgment text. The conversion endpoint is what the SPA itself calls client-side to render the document.
• Very recently published documents (seen: within ~2-4 days) can 204 on the conversion endpoint — the HTML conversion lags the search index. echrjudgment reports this explicitly (found: false, reason: "notyetconverted") rather than returning empty text silently; echrsearch's conclusion field is populated immediately regardless.
• French-language-only documents (doctype starting HF) frequently 204 on the same endpoint even when old; echr_search defaults to languageisocode=ENG and excludes doctype=PR (press-release stubs, which otherwise leak into every collection filter) to route around both traps.
• respondent takes an ISO3166-1 alpha-3 code (FRA, GBR, ...); the tool also accepts common country names via a small lookup table covering all 46 member states plus Russia (expelled 2022, but its pre-expulsion case law is still in HUDOC and still gets asked about).

Quick Start

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

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

What this endpoint actually serves

tools/list at https://gateway.pipeworx.io/hudoc/mcp returns the tools in the table above plus the shared Pipeworx meta-tools — askpipeworx, discovertools, 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:

Shortened. The full README is on GitHub.