sweden-parliament-mcp-server

---

Overview

Swedish parliamentary records from Riksdagen Open Data. Find proposals and committee reports, follow published decisions and main votes, read speeches, explore member assignments, and retrieve consolidated SFS statutes. Runs as a stdio process or a local Streamable HTTP server; no API key required.

Source: The Swedish Parliament. This independent service is not operated or endorsed by Sveriges riksdag. Swedish titles, labels, decisions, and text retain their source wording; English labels appear only where the source supplies them.

Tools

Tool	Description
riksdag_list_reference	Browse official document types, committees, parties, constituencies, sessions, and identifier and coverage guidance
riksdag_search_documents	Search parliamentary documents and consolidated SFS statutes by words, type, session, committee, party, or dates
riksdag_get_document	Read publication metadata, decisions, reservations, activities, links, and paginated text
riksdag_search_speeches	Find speeches within a session by debate heading, member, or party; retrieve full text through get mode
riksdag_get_votes	Read report decisions with party main-vote totals or one member's session main votes
riksdag_search_members	Find register entries by name, party, constituency, and source service classification
riksdag_get_member	Read an exact member IID and its published assignment history

Capability reference

riksdag_list_reference _tool

• topic: document_types, committees, parties, constituencies, speech_types, sessions, identifiers, or coverage
• Dated offline snapshot with Swedish labels, available source English labels, source links, and referenceVerifiedAt; document subtypes remain separate from parent codes, and the session catalog can lag current sessions
• offset pagination; limit defaults to 20 and is clamped to 100, with totalCount and complete nextCall arguments

---

riksdag_search_documents _tool

• Optional query, documentType, session, committee, party, dateFrom, and dateTo; dates use YYYY-MM-DD, and codes or exact reference labels are accepted
• Remote page starts at 1; pageSize defaults to 10 and is clamped to 50. Every fetched row is returned, with source ordering, available totalCount, and nextCall
• Use documentType: "sfs" for consolidated statutes; narrow slow full-text searches with session or documentType

---

riksdag_get_document _tool

• Required documentId from search, or an unambiguous SFS identifier such as 1974:152; missing records return found: false
• Omit sections for metadata, the complete outline, and the first text page; select metadata, decisions, reservations, activities, properties, attachments, references, text, or source_html
• Text and HTML use Unicode code-point offsets (textOffset / htmlOffset); lengths default to 1,000 and explicit textLength / htmlLength values are honored. Array sections use sectionOffset / sectionLimit (default 20, maximum applied 100)
• Follow nextTextCall, nextHtmlCall, or each sectionPages[].nextCall, retaining revision; changed records fail with revision_changed. Attachments are links, not downloaded files
• SFS versionKind: "consolidated" describes published consolidated text, which may include future-effective provisions; it is not an as-enacted or point-in-time legal version

---

riksdag_search_speeches _tool

• mode: "search" (default) requires session; optional memberId, party, and speechType: "all" | "speech" narrow the selection, with speech excluding replies
• headingContains matches all tokens against headings and subheadings, ignoring case and diacritics; it never searches speech body text
• Local offset pagination; limit defaults to 5 and is clamped to 20. textCount defaults to 1, is clamped to 5, and accepts 0 for metadata only
• mode: "get" requires a retrieval speechId copied from search, not sourceSpeechId; get mode rejects search filters. textOffset and optional textLength read the whole published body through nextTextCall; the default text page is 1,000 code points
• textState is included, not_requested, unavailable, or failed; transient hydration failures include recovery details. Retain revision on search or text continuations

---

riksdag_get_votes _tool

• Required mode and session; mode: "report" takes report (designation or session:designation) or a report documentId, while mode: "member" requires memberId and permits party
• Report mode returns authoritative decisions and complete party groups. Select one point, or use pointOffset / pointLimit (default 5, maximum applied 10); member mode uses offset / limit (default 20, maximum applied 500)
• Published main votes only, not every ballot round. rollCallState is available, acclamation, not_published, or unknown; acclamation requires explicit decision evidence, never merely empty vote rows
• Missing aggregate counts remain unknown; additional source vote labels stay in otherVotes. Follow nextCall with revision for remaining whole points or member rows

---

riksdag_search_members _tool

• Optional nameContains matches all tokens ignoring case and diacritics; party and constituency accept codes or exact reference labels
• sitting: "all" | "serving" | "not_serving" defaults to all; this filters source service classification, not verified sitting or elected membership. Historical entries and ministers can appear, and unknown classifications remain in all
• Local offset pagination; limit defaults to 20 and is clamped to 100. Returns candidateCount, exact filtered totalCount, revision, and nextCall
• Keep leading zeros in memberId. Some European Parliament entries have no IID and return detailAvailable: false; their sourceRecordId is a different identifier and cannot be used with riksdag_get_member

---

riksdag_get_member _tool

• Required numeric-string memberId from search, including leading zeros; missing entries return found: false with guidance
• Published register metadata and assignments in source order; assignmentOffset / assignmentLimit default to 0 / 20, with a maximum applied limit of 100
• assignmentTotal, nextCall, and revision allow complete traversal. Assignment dates retain source precision and do not establish historical sitting status; contact details and photographs are excluded

Features

Built on @cyanheads/mcp-ts-core: stdio and Streamable HTTP transports, pluggable auth (none / jwt / oauth), swappable storage (in-memory, filesystem, Supabase, Cloudflare KV/R2/D1), structured logging with optional OpenTelemetry tracing.

Riksdag-specific:

• Keyless access to official document, speech, vote, and member endpoints, plus dated official reference tables
• Document text as canonical Markdown or SFS source text, with separately selectable original HTML
• Shared upstream concurrency, bounded public-response caching, and one 55-second deadline across queueing, retries, and text hydration
• Historical coverage varies by record type; speeches and votes start in 1993/94, with gaps in older vote data

Agent-friendly output:

• Source attribution, retrieval time, effective filters, and source warnings on both content[] and structuredContent
• Complete continuation calls for rows, sections, and text, with revision checks on local continuations; a changing remote document search is not a frozen snapshot
• Explicit misses, unavailable text, unknown counts, and source service classifications; candidate saturation fails as candidate_limit_reached instead of returning an incomplete traversal

Getting started

Add the following to your MCP client configuration file.

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

Or with npx (no Bun required):

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

Or with Docker:

{
  "mcpServers": {
    "sweden-parliament-mcp-server": {
      "type": "stdio",
      "command": "docker",
      "args": ["run", "-i", "--rm", "-e", "MCP_TRANSPORT_TYPE=stdio", "ghcr.io/cyanheads/sweden-parliament-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+).
• Riksdagen Open Data requires no API key or account.

Installation

1. Clone the repository:

git clone https://github.com/cyanheads/sweden-parliament-mcp-server.git

2. Navigate into the directory:

cd sweden-parliament-mcp-server

3. Install dependencies:

bun install

4. Configure environment:

cp .env.example .env
# edit .env for optional transport, authentication, or logging overrides

Configuration

Variable	Description	Default
MCP_TRANSPORT_TYPE	Transport: stdio or http.	stdio
MCP_HTTP_PORT	HTTP server port.	3010
MCP_SESSION_MODE	HTTP session mode: stateless, stateful, or auto.	stateless
MCP_AUTH_MODE	Authentication: none, jwt, or oauth.	none
MCP_LOG_LEVEL	Log level (debug, info, warning, error, etc.).	info
LOGS_DIR	Directory for log files (Node.js only).	<app-root>/logs
STORAGE_PROVIDER_TYPE	Storage backend: in-memory, filesystem, supabase, cloudflare-kv/r2/d1.	in-memory
OTEL_ENABLED	Enable OpenTelemetry.	false

See .env.example for every server setting and the common framework overrides.

Running the server

Local development

• Build and run the production version:

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

• Run checks and tests:

  bun run devcheck  # Lints, formats, type-checks, and more
  bun run test      # Runs the test suite
  

Project structure

Directory	Purpose
src/index.ts	Registers seven tools, initializes the service, and selects stateless HTTP sessions.
src/mcp-server/tools/definitions	Seven read-only tool definitions; no resources or prompts.
src/services/riksdag	Riksdagen endpoint parsing, reference tables, transport, cache, text conversion, and continuation.
tests/	Unit and integration tests, mirroring the src/ structure.

See docs/tree.md for the full file tree.

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 logging, ctx.state for storage
• Register new tools and resources in the createApp() arrays in src/index.ts
• Wrap external API calls: validate raw → normalize to domain type → return output schema; never fabricate missing fields

Contributing

Issues are welcome. Run checks and tests before submitting:

bun run devcheck
bun run test

License

Apache-2.0 — see LICENSE for details.