Tools

Seven tools covering EU legal research — document discovery, content retrieval, citation resolution, case law, relationship graph traversal, EuroVoc thesaurus lookup, and raw SPARQL access:

eurlex_search_documents

Search EU legislation, treaties, preparatory acts, and more across the 2.7M+ work CELLAR corpus.

• Keyword search across work titles and CELEX string patterns
• Filter by document category (REG, DIR, DEC, TREATY, and more), with each category expanding to its reviewed CELLAR authority family
• Optionally include consolidated texts whose basic act belongs to the selected category
• Date range filtering (date_from, date_to)
• EuroVoc concept filtering — use eurlex_browse_subjects first to resolve concept IDs
• Filter to in-force acts only
• Pagination via offset and configurable limit (max 100)
• Returns CELEX numbers, work URIs, document types, and dates for chaining into eurlex_get_document

eurlex_get_document

Fetch the notice and full text of an EU legal act.

• Accepts CELEX numbers (e.g., 32016R0679), ELI URIs, or CELLAR work URIs
• Returns structured metadata: title, date, document type, author institution, legal basis, EuroVoc subjects, in-force flag
• Full text in HTML (default), Markdown, or Formex4 XML — format: "markdown" converts the act body to clean Markdown server-side (recitals and numbered points as readable text, genuine data tables as GFM)
• Content shaping for large acts: ordinary offset-based "paged" windows and "full" windows are capped at 100,000 characters; content_mode "paged" (default) returns the requested character window (offset + limit), while "full" returns the first window from offset zero. Both include content_chars_total, content_offset, content_chars_returned, and has_more, so repeated paged calls can reconstruct the complete body without loss; "metadata_only" skips the body, while structural outline/selection behavior is unchanged
• Navigate structure instead of raw offsets: outline: true returns the act's chapters, articles, annexes, and recitals as a heading list (each with its character offset), and select (e.g. { articles: "1,5,17" }) returns just those sections' text — degrading cleanly to the paging floor for acts with no detectable structure (e.g. case law)
• Supports the 24 EUR-Lex language codes case-insensitively, normalizes them to uppercase, defaults to English, and automatically falls back to English when a requested translation is unavailable
• Preserves the exact shaped body in structured output; HTML and XML are shown literally in the text response inside a source-safe tilde fence, while Markdown remains rendered Markdown
• Older acts and some CJEU judgments may lack English translations

eurlex_lookup_celex

Resolve EU legal identifiers to canonical CELLAR works.

• Accepts CELEX numbers and ELI URIs
• Auto-detects format with identifier_type: "auto" (default); set explicitly when auto-detection fails
• Returns work URI, confirmed CELEX number, document type, and date — the prerequisite step before eurlex_get_document or eurlex_get_relations

eurlex_get_cases

Search CJEU and General Court case law.

• Case-specific search: case number, keyword, court (CJEU or GC), and case type (judgment, order, ag_opinion)
• Date range filtering
• Primary records only by default — derivative information notices, abstracts, and summaries are excluded; set include_derivative to include them
• Returns case identifier, court, date, document type, and parties
• Distinct from eurlex_search_documents — case law (CELEX sector 6) has its own search parameters and practitioner workflows

eurlex_get_relations

Traverse the CELLAR relationship graph for a given work.

• Amendment chain (what amends it, what it amends)
• Consolidated versions (the current in-force text)
• Legal basis
• Citation network (cdm:work_cites_work in both directions)
• National transposition measures
• Filter to specific relation types or retrieve all at once
• Returns one-hop relations; multi-hop traversal requires multiple calls or eurlex_query_sparql

eurlex_browse_subjects

Resolve human-readable terms to EuroVoc concept IDs.

• Full-text search across the multilingual EuroVoc thesaurus
• Returns concept URI, preferred label, concept code, and broader/narrower hierarchy hints
• Supports all EU official languages; defaults to English
• Required before using the eurovoc_concept filter in eurlex_search_documents

Resources and prompts

All resource data is also reachable via tools. Resources provide stable-URI injectable context for agents that support MCP resources.

Features

Built on @cyanheads/mcp-ts-core:

• Declarative tool, resource, and prompt definitions — single file per primitive, framework handles registration and validation
• Unified error handling — handlers throw, framework catches, classifies, and formats
• Pluggable auth: none, jwt, oauth
• Swappable storage backends: in-memory, filesystem, Supabase, Cloudflare KV/R2/D1
• Structured logging with optional OpenTelemetry tracing
• STDIO and Streamable HTTP transports
• MCP 2026-07-28 support, exact tool-input validation, and structured error envelopes clients can act on

EUR-Lex-specific:

• No API key required — both CELLAR SPARQL and EUR-Lex REST content endpoints are publicly accessible
• CellarSparqlService POSTs application/x-www-form-urlencoded SPARQL with CDM prefix declarations built in; server-side LIMIT enforcement (max 100) prevents Virtuoso timeout abuse
• EurLexContentService fetches act text from the CELLAR content-negotiation resolver (/resource/celex/{CELEX} with Accept / Accept-Language headers); HTML and Formex4 XML pass through, Markdown is converted server-side from the HTML body
• Virtuoso error classification: HTTP 200 with Virtuoso 37000 Error body is parsed and re-raised as ServiceUnavailable (transient/timeout) or InvalidParams (syntax error)
• Language fallback on document fetch: if the requested language is unavailable, retries with English and reports requested/effective language parity; ordinary absence after both attempts remains a successful metadata response with content_status and a typed unavailability cause, while a WAF challenge remains a typed content_challenge error
• Typed error contracts on every tool — structured reason codes let agents branch on outcomes without parsing text

Agent-friendly output:

• EuroVoc prerequisite guidance in server-level instructions — agents are directed to eurlex_browse_subjects before attempting concept-filtered searches
• eurlex_lookup_celex surfaces CELEX confirmation and work existence upfront, preventing downstream errors in document or relation fetches
• content_status, content_unavailability_reason, and requested/effective language fields let agents distinguish skipped, available, absent, upstream-failed, and incomplete multipart content without string parsing
• Relationship graph output carries relation type labels alongside CELLAR URIs and resolved CELEX numbers for human-readable downstream use

Getting started

Public Hosted Instance

A public instance is available at https://eur-lex.caseyjhand.com/mcp — no installation required. Point any MCP client at it via Streamable HTTP:

{
  "mcpServers": {
    "eur-lex-mcp-server": {
      "type": "streamable-http",
      "url": "https://eur-lex.caseyjhand.com/mcp"
    }
  }
}

Self-Hosted / Local

Add the following to your MCP client configuration file. No API key is required.

{
  "mcpServers": {
    "eur-lex-mcp-server": {
      "type": "stdio",
      "command": "bunx",
      "args": ["@cyanheads/eur-lex-mcp-server@latest"],
      "env": {
        "MCP_TRANSPORT_TYPE": "stdio",
        "MCP_LOG_LEVEL": "info"
      }
    }
  }
}

Or with npx (no Bun required):

{
  "mcpServers": {
    "eur-lex-mcp-server": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@cyanheads/eur-lex-mcp-server@latest"],
      "env": {
        "MCP_TRANSPORT_TYPE": "stdio",
        "MCP_LOG_LEVEL": "info"
      }
    }
  }
}

Or with Docker:

{
  "mcpServers": {
    "eur-lex-mcp-server": {
      "type": "stdio",
      "command": "docker",
      "args": [
        "run", "-i", "--rm",
        "-e", "MCP_TRANSPORT_TYPE=stdio",
        "ghcr.io/cyanheads/eur-lex-mcp-server:latest"
      ]
    }
  }
}

For Streamable HTTP, set the transport and start the server:

MCP_TRANSPORT_TYPE=http MCP_HTTP_PORT=3010 bun run start:http
# Server listens at http://localhost:3010/mcp

Prerequisites

• Bun v1.4.0 or higher (or Node.js v24+).
• No API key needed — EUR-Lex and CELLAR are publicly accessible.

Installation

1. Clone the repository:

git clone https://github.com/cyanheads/eur-lex-mcp-server.git

2. Navigate into the directory:

cd eur-lex-mcp-server

3. Install dependencies:

bun install

4. Configure environment (optional):

cp .env.example .env
# All server-specific vars have sensible defaults — no required vars

Configuration

All configuration is validated at startup via Zod schemas in src/config/server-config.ts.

See .env.example for the full list of optional overrides.

Running the server

Local development

• Build and run:

  sh

  Copy

  # One-time build
  bun run rebuild
  
  # Run the built server
  bun run start:stdio
  # or
  bun run start:http
  

• Run checks and tests:

  sh

  Copy

  bun run devcheck   # Lint, format, typecheck, security
  bun run test       # Vitest test suite
  bun run lint:mcp   # Validate MCP definitions against spec
  

Docker

docker build -t eur-lex-mcp-server .
docker run --rm -p 3010:3010 eur-lex-mcp-server

The Dockerfile defaults to HTTP transport, stateless session mode, and logs to /var/log/eur-lex-mcp-server. OpenTelemetry peer dependencies are installed by default — build with --build-arg OTEL_ENABLED=false to omit them.

Project structure

Development guide

See CLAUDE.md for development guidelines and architectural rules. The short version:

• Handlers throw, framework catches — no try/catch in tool logic
• Use ctx.log for request-scoped logging, ctx.state for tenant-scoped storage
• Register new tools and resources via the barrels in src/mcp-server/*/definitions/index.ts
• Wrap external API calls: validate raw → normalize to domain type → return output schema; never fabricate missing fields

Contributing

Issues and pull requests are welcome. Run checks and tests before submitting:

bun run devcheck
bun run test

License

Apache-2.0 — see LICENSE for details.