parlament-mcp
Summary: A read-only MCP server that connects AI models to the Swiss Federal Parliament (Curia Vista) for searching motions, members, votes, sessions and verbatim debate transcripts — no API key required.
Search parliamentary businesses (motions, interpellations, postulates, initiatives) by keyword + second keyword, type, status, council or submission date.
Get full details of one business by ID: motion text, submitted text, initial situation, Federal Council response, proceedings and status.
Find councillors by canton, party, council or last name (active-only by default).
Retrieve votes with
meaning_yes/meaning_noso a Ja/Nein can be interpreted, filtered by keyword or session.List recent sessions with IDs and dates to use as filters for votes and transcripts.
Search debate transcripts (Amtliches Bulletin) for short, citable excerpts with AB citation, source URL and real spoken language — filter by speaker, session, business number, council or date window.
Fetch the verbatim full text of a single speech by transcript ID, capped at
max_charsand paginated viais_excerpt/next_offset— never summarised.Safe by design: all tools read-only, idempotent, open-world; paginated result caps; every response carries source, license, provenance and match type. Coverage: businesses since 1978, transcripts from 1999-12-06.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@parlament-mcp Finde alle hängigen Motionen zum Thema KI in der Bildung"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
🏛️ parlament-mcp
Part of the Swiss Public Data MCP Portfolio – connecting AI models to Swiss public data sources.
Note: This server covers the federal level (Curia Vista). This repo additionally hosts a self-contained subproject under
openparldata-mcp/— the subnational counterpart for the 26 cantons and ~70 municipal parliaments (OpenParlData.ch). The two are independent servers.
An MCP server that connects AI models to the Swiss Federal Parliament via the
Curia Vista OData API (ws.parlament.ch).
Access motions, interpellations, votes, members, sessions, and the verbatim
debate transcripts of the Amtliches Bulletin – with no API key required
(Phase 1 – No-Auth-First).
🎯 Anchor Demo Queries
Metadata layer:
"Welche Vorstösse zu KI in der Schule sind hängig?" →
parlament_search_business(keyword="KI", keyword2="Schule", status="Eingereicht")
Verbatim transcripts (Amtliches Bulletin):
"What did National Councillor Munz say in the 2024 spring session about the Volksschule? Give me the exact wording with a correct AB citation." →
parlament_search_transcripts(speaker_name="Munz", session_id=5202, keyword="Volksschule")→ thenparlament_get_transcript(transcript_id=…)for the full wording.Returns short, citable excerpts (
AB 2024 N, 2024-03-13, Munz Martina) with a stable source URL; the verbatim text is fetched on demand, never in bulk.
Perfect for the KI-Fachgruppe Stadtverwaltung Zürich: find pending motions on AI in education, or quote what was actually said in the chamber – instantly.
Related MCP server: lobbywatch-mcp
🔧 Tools
Tool | Description |
| Search Vorstösse by keyword, type, status, council, date |
| Full details of a single business (texts, FC response) |
| Find councillors by canton (e.g. ZH), party, council |
| Parliamentary votes with Ja/Nein meaning |
| List recent sessions with IDs for follow-up queries |
| Search debate transcripts → citable excerpts with AB citation + source URL (speaker / session / business / date filters) |
| Fetch the verbatim full text of a single speech by ID (capped, paginated, |
🏗️ Architecture
┌──────────────────────────────────┐
│ MCP Host (Claude Desktop / │
│ Claude API / IDE) │
└─────────────┬─────────────────────┘
│ MCP Protocol (JSON-RPC 2.0)
│ Transport: stdio (local) / SSE (cloud)
┌─────────────▼─────────────────────┐
│ parlament-mcp │
│ FastMCP · Python · Pydantic v2 │
│ │
│ ┌── metadata layer (server.py) ──┐
│ │ search_business · get_business │
│ │ search_members · get_votes │
│ │ get_sessions │
│ └─────────────────────────────────┘
│ ┌── transcript layer ────────────┐ ← separate module (transcripts.py)
│ │ search_transcripts (excerpts) │ · Language='DE' dedups editions
│ │ get_transcript (verbatim) │ · Type=1 = real speeches only
│ └─────────────────────────────────┘ · retry + 45s read timeout
└─────────────┬─────────────────────┘
│ HTTPS / OData v3
┌─────────────▼─────────────────────┐
│ ws.parlament.ch / odata.svc │
│ Curia Vista – No Auth Required │
│ │
│ Business · Vote · MemberCouncil │ metadata path
│ Session ─< Meeting ─< Subject ─< Transcript transcript path
└───────────────────────────────────┘🚀 Installation
Claude Desktop (stdio)
Add to ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"parlament": {
"command": "uvx",
"args": ["parlament-mcp"]
}
}
}Local development
git clone https://github.com/malkreide/parlament-mcp
cd parlament-mcp
pip install -e .
python -m parlament_mcp.serverCloud / Railway (SSE)
MCP_TRANSPORT=sse MCP_HOST=0.0.0.0 PORT=8080 python -m parlament_mcp.server
# SSE endpoint: http://your-host:8080/sseNetwork binding
By default the server binds to 127.0.0.1 (localhost only). Set
MCP_HOST=0.0.0.0 only inside a container/cloud context (Docker, Railway,
Render, Kubernetes). Never bind to 0.0.0.0 on a local dev machine – it exposes
the server to your local network (NeighborJack risk); the server logs a warning
if you do so outside a detected container.
Transport is selected via MCP_TRANSPORT (stdio default, or sse /
streamable-http); --http is kept as an alias for streamable-http.
Authentication (optional)
The HTTP transport is open by default (public read-only data). To require a
bearer token, serve via the CORS/auth app factory and set MCP_BEARER_TOKENS:
MCP_BEARER_TOKENS="alice:tok_abc,bob:tok_def" MCP_ALLOWED_ORIGINS="https://claude.ai" \
uvicorn parlament_mcp.server:create_http_app --factory --host 0.0.0.0 --port 8080Each request then needs Authorization: Bearer <token>; identity comes from the
validated token, not a session header (see docs/security.md).
Docker
docker compose up --build # binds 127.0.0.1:8080 only
# or build the hardened image directly (non-root, read-only FS):
docker build -t parlament-mcp .Kubernetes manifests (hardened securityContext, resource limits, egress
NetworkPolicy, Mcp-Session-Id sticky routing) live in deploy/k8s/;
an HAProxy stick-table example is in deploy/haproxy.cfg.
🔗 Synergies
Partner Server | Combination |
Federal ↔ subnational — same question across cantons & municipalities | |
Law text ↔ parliamentary debate that created it | |
City policy ↔ cantonal/federal motions | |
Data backing ↔ motions citing statistics |
Power query example:
"Zeige mir alle Zürcher Motionen zu KI in der Bildung
und verlinke die relevanten Bundesgesetze aus fedlex-mcp."📊 Data Source
Authentication: None (Phase 1 – No-Auth-First)
Protocol: OData v3 / JSON
Coverage: All parliamentary businesses since 1978; votes since ~2000. Structured verbatim transcripts (Amtliches Bulletin) from 1999-12-06 onward (earlier years 1891–1999 exist only as archive scans — see Known Limitations).
Update cycle: Real-time (official government data)
Copyright — verbatim quotation is allowed
Official proceedings of Swiss authorities are excluded from copyright under Art. 5 para. 1 lit. a URG (Swiss Copyright Act). The verbatim wording of parliamentary debates in the Amtliches Bulletin may therefore be reproduced and quoted freely — which is exactly what the transcript tools return: the wording itself, never a summary standing in for it.
📜 Data sources & licenses
Source | License | Attribution |
Curia Vista (ws.parlament.ch) | CC BY 4.0 | © Schweizer Parlament, CC BY 4.0 |
Every tool returns a typed structured response (FastMCP exposes the output
schema) carrying source, license, provenance, match_type and count
alongside typed results. Data is passed through unmodified.
🧭 Phase
This server is in Phase 1 — Read-only Wrapper (all tools readOnlyHint: true,
no writes). The full phase model and transition criteria are in
docs/roadmap.md.
🔖 MCP Protocol Version
This server speaks two protocol eras over the same endpoint. The client's first request on a connection decides which one applies; a later claim from the other era is refused.
Era | Revision | Who reaches it |
|
| What today's clients speak. The server answers with the revision asked for, or with the |
Per-request envelope |
| A request carrying the |
Both revisions are pinned in
tests/test_protocol_version.py and asserted
against the installed SDK, so a Dependabot bump of mcp cannot move either one
silently. The same file also measures the negotiation: one real client
connection per era against the registered server, reading the revision off the
connection instead of inferring it from a constant.
In the 2026-07-28 era there is no handshake to carry the server's identity;
the SDK stamps io.modelcontextprotocol/serverInfo into the result _meta
instead. The server passes its package version there
(tests/test_server_info.py), so both eras report
the same version.
Note that the SDK's LATEST_PROTOCOL_VERSION is an alias for the modern
era, not for the handshake era — pinning against it alone would leave the era
that current clients actually negotiate free to drift.
Update policy. When the gate fails, do not edit the constant blindly: read
the spec changelog between the two revisions, verify the server still behaves,
then move the constant, this section, README.de.md and
CHANGELOG.md together.
🧱 MCP primitives
Phase 1 uses Tools only — rationale and the Phase-2 Resources plan are in
docs/adr/ADR-003-mcp-primitives.md.
🏷️ Tool annotations
All tools declare explicit annotations consistent with their behaviour:
Tool | readOnly | destructive | idempotent | openWorld |
| ✅ | — | ✅ | ✅ |
| ✅ | — | ✅ | ✅ |
| ✅ | — | ✅ | ✅ |
| ✅ | — | ✅ | ✅ |
| ✅ | — | ✅ | ✅ |
| ✅ | — | ✅ | ✅ |
| ✅ | — | ✅ | ✅ |
📈 Observability
Structured JSON logs go to stderr (stdout stays reserved for the stdio
protocol). OpenTelemetry tracing wraps each tool call and auto-instruments
outgoing HTTP; set OTEL_EXPORTER_OTLP_ENDPOINT (with the otel-export extra)
to ship spans. See docs/security.md for the full security
posture (Lethal-Trifecta assessment, egress allow-list, gateway hardening).
🛡️ Safety & Limits
Aspect | Details |
Access | Read-only ( |
Personal data | Parliamentary businesses are public record by law (BGÖ). No private data is accessed or stored. |
Rate limits | Built-in per-query caps: max. 100 results (businesses/members), 50 (votes), 30 (transcript search), 10 (sessions). Transcript full text is capped per call and paginated — you never pull a whole session by accident. |
Timeout | 20 seconds per metadata call; 45 seconds for transcript reads (verbatim search is heavier) |
Authentication | No API keys required — Curia Vista is publicly accessible |
Data source | Official Swiss federal government data (Schweizerische Parlamentsdienste) |
Terms of Service | Subject to ToS of ws.parlament.ch — Schweizerische Parlamentsdienste |
Known Limitations
General
OData
substringof()filter is case-sensitive for some fields.Session names may be
nullin the API for very recent sessions – use session ID.
Transcripts (Amtliches Bulletin) — verified live 2026-07-19:
Temporal coverage: from 1999-12-06 only. The structured
Transcriptentity reaches back to Dec 1999. Debates from 1891–1999 exist only as scanned archive documents (Bundesarchiv / Amtsdruckschriften) and are not connected here (no OCR in scope). A query whose whole date window predates coverage returns an explanatory error, not an empty result.No page number in the source. The API carries no page/column field, so the classic
AB <year> N <page>form cannot be built. We emit a stable, verifiable substitute —AB <year> <N|S>, <date>, <speaker>— plus the authoritativesource_url(SubjectId) and thetranscript_id. This is a documented, honest trade-off, not an omission.Language behaviour (important).
Languageis the edition, not the spoken language. The tools filterLanguage eq 'DE'purely to deduplicate the three byte-identical editions (DE/FR/IT) down to one copy. Every speech is returned in its original wording; a French or Italian speech is not hidden — its real language is reported in thelanguagefield (de/fr/it). The response states this vialanguage_note.Truncation is explicit, never silent. Search returns short excerpts (
snippet, ~320 chars) withis_excerpt,total_length_charsand a hint to fetch the full text.parlament_get_transcriptcaps output atmax_chars; when it truncates it setsis_excerpt=Trueand returns anext_offsetto continue.Verbatim only, never summarised. The wording is the product; the tools never substitute a summary for the actual text.
Latency: a free-text
keywordcombined with aspeaker_nameis the slowest path (~40 s). Add asession_id,business_numberor date window to keep reads around 1–2 s.
🧪 Testing
pip install -e ".[dev]"
# Unit + mocked integration tests (no network), as run in CI:
PYTHONPATH=src pytest tests/ -m "not live"
# Include live tests against the real ws.parlament.ch API:
PYTHONPATH=src pytest tests/ -m liveHTTP is mocked with respx; network-dependent tests are marked
@pytest.mark.live and excluded from CI via -m "not live". Tool definitions are
pinned in tool-hashes.json (python -m parlament_mcp.tool_hashes --check).
Contributing
See CONTRIBUTING.md.
Security
See SECURITY.md for the security policy and posture (vulnerability reporting, Lethal-Trifecta assessment, accepted risks).
License
MIT © Hayal Oezkan — see LICENSE
Author
Hayal Oezkan · github.com/malkreide
Installation
Run via uv's uvx — no clone or manual install needed. Add to your MCP client config (mcpServers for Claude Desktop, Cursor and Windsurf; use a top-level servers key for VS Code in .vscode/mcp.json):
{
"mcpServers": {
"parlament-mcp": {
"command": "uvx",
"args": [
"parlament-mcp"
]
}
}
}Available Tools
7 toolsparlament_get_businessARead-onlyIdempotent
Vollständige Details eines parlamentarischen Vorstosses nach Curia Vista ID abrufen.
<use_case>Nach einer Suche verwenden, um vollständige Informationen inkl. Ausgangslage, Vorstosstext und Antwort des Bundesrats zu erhalten.</use_case>
<important_notes>Benötigt die numerische Geschäfts-ID (aus
parlament_search_business). found=false bei unbekannter ID.</important_notes>
Inputs: params (required). Returns: id, url, tags, type, +17 more
parlament_get_sessionsARead-onlyIdempotent
Aktuelle parlamentarische Sessionen mit Daten auflisten.
<use_case>Session-IDs aus dieser Liste zum Filtern von Abstimmungen oder Transkripten verwenden.</use_case>
<important_notes>Session-Namen können für sehr aktuelle Sessionen null
sein – dann die Session-ID verwenden.</important_notes>
Inputs: params (required). Returns: note, count, offset, source, +5 more
parlament_get_transcriptARead-onlyIdempotent
Den vollen Wortlaut eines einzelnen Votums nach Transkript-ID abrufen.
<use_case>Nach einer Suche mit parlament_search_transcripts den kompletten Wortlaut eines konkreten Votums holen – zitierfähig, mit stabiler URL.</use_case>
<important_notes>Ausgabe ist gedeckelt (max_chars); bei Kürzung ist
is_excerpt=True gesetzt und note/next_offset erklären, wie die
Fortsetzung zu laden ist. Kein stilles Kürzen. Der Wortlaut wird nie durch
eine Zusammenfassung ersetzt.</important_notes>
Inputs: params (required). Returns: Volltext eines *einzelnen* Votums – gedeckelt, mit Kürzungs-Kennzeichnung.
parlament_get_votesBRead-onlyIdempotent
Parlamentarische Abstimmungen (im Rat) mit Ja/Nein-Bedeutung abrufen.
<use_case>Zeigt, wie der Rat über Themen wie KI-Regulierung, Bildungsfinanzierung oder Digitalisierungsprojekte abgestimmt hat.</use_case>
<important_notes>meaning_yes/meaning_no erklären, was ein Ja/Nein im
konkreten Geschäft bedeutet – wichtig zur korrekten Interpretation.</important_notes>
Inputs: params (required). Returns: note, count, offset, source, +5 more
parlament_search_businessARead-onlyIdempotent
Parlamentarische Vorstösse suchen (Motionen, Interpellationen, Postulate usw.).
Durchsucht Curia Vista Geschäftsdaten von ws.parlament.ch.
<use_case>Politische Recherche zu Bildungs-, Datenschutz- oder Verwaltungsthemen; hängige Vorstösse zu KI in der Bildung, Digitalisierungsinitiativen oder beliebigen Politikthemen finden.</use_case>
<important_notes>Titel-Suche via OData substringof() (gross-/klein-sensitiv).
Maximal 100 Treffer pro Aufruf; mit offset paginieren.</important_notes>
keyword='KI', keyword2='Schule', status='Eingereicht'
Inputs: params (required). Returns: note, count, offset, source, +5 more
parlament_search_membersARead-onlyIdempotent
National- und Ständeräte suchen.
<use_case>Alle Zürcher Ratsmitglieder ('ZH') oder Mitglieder einer bestimmten Partei finden. Synergie: mit parlament_search_business kombinieren, um Urheber von Vorstössen zu identifizieren.</use_case>
<important_notes>active_only=True (Default) liefert nur amtierende
Mitglieder. Kanton als 2-Buchstaben-Kürzel.</important_notes>
Inputs: params (required). Returns: note, count, offset, source, +5 more
parlament_search_transcriptsARead-onlyIdempotent
Wörtliche Wortmeldungen aus den Ratsdebatten durchsuchen (Amtliches Bulletin).
Liefert kurze, zitierfähige Auszüge (kein Volltext) mit korrekter
AB-Zitation und stabiler Quell-URL. Für den Wortlaut eines einzelnen Votums
danach parlament_get_transcript(transcript_id=…) verwenden.
<use_case>«Was hat Nationalrätin X in der Frühjahrssession 2024 zur Volksschule gesagt?» – Sprecher, Session, Rat, Geschäft oder Datumsfenster kombinieren. Synergie mit fedlex-mcp: vom Gesetzestext zur Debatte.</use_case>
<important_notes>Nur echte Wortmeldungen (keine Abstimmungszeilen). Der
Language-Filter dedupliziert die Editionen und blendet keine
französisch-/italienischsprachigen Voten aus – die reale Sprache steht in
language. Abdeckung ab 1999-12-06. Für beste Latenz session_id,
business_number oder ein Datumsfenster mit einem freien keyword
kombinieren.</important_notes>
speaker_name='Munz', session_id=5202, keyword='Volksschule'
Inputs: params (required). Returns: note, count, offset, source, +10 more
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 7 tools
Each tool targets a distinct resource and action: search vs. get, and business/transcripts/members/votes/sessions are clearly separated. The two search tools return different entity types, and the two get tools are distinguished by ID and purpose.
All tool names follow a consistent verb_noun pattern with 'search_' for queries and 'get_' for retrievals. No mixed conventions or vague verbs.
Seven tools cover the core parliamentary data domain without bloat. Each tool has a clear, non-redundant role, and the count is within the ideal 3-15 range.
The set covers search and retrieval for proposals, transcripts, members, votes, and sessions, which are the main entities. Minor gaps include no direct get_member by ID and no committee data, but agents can work around these with search_members and get_votes.
Maintenance
Related MCP Connectors
MCP server for Brazilian Federal Senate open data (legislative, administrative, e-Cidadania).
An MCP server that provides congressional transcripts
MCP server unifying ERPs, CRMs, APIs and knowledge base for Claude, ChatGPT and Gemini.
Hosted MCP server for live public-data APIs and Skills for AI agents.
Related MCP Servers
AlicenseAqualityDmaintenanceMCP server for Portuguese Parliament open data, enabling AI agents to access legislative initiatives, deputies, plenary votes, petitions, and parliamentary committees.12MIT- AlicenseAqualityBmaintenanceAn MCP server that connects AI models to the Lobbywatch.ch database, providing access to Swiss parliamentarians' conflicts of interest, lobby groups, access badges, and transparency scores.851 PyPIMIT
- AlicenseAqualityBmaintenanceMCP server connecting AI models to Swiss Federal Food Safety and Veterinary Office open data, enabling queries about food recalls, animal disease surveillance, food control results, and more.1134 PyPIMIT
- AlicenseBqualityBmaintenanceAn MCP server providing AI-powered access to Open Data from the City of Zurich, enabling queries to 900+ datasets, real-time environmental and mobility data, geodata, parliamentary proceedings, and more.2666 PyPI9MIT