Swiss Geodata
БесплатноНе проверенMCP server that connects AI models to Swiss federal geodata, offering tools for spatial queries like layer discovery, coordinate identification, building zones,
Описание
MCP server that connects AI models to Swiss federal geodata, offering tools for spatial queries like layer discovery, coordinate identification, building zones, and terrain heights.
README
⚠️ DEPRECATED — use swisstopo-mcp instead
This server has been consolidated into swisstopo-mcp and is no longer maintained. Every capability it offered now exists there, including the one it used to have exclusively — the official REFRAME coordinate conversion.
Why: both servers wrapped the same
api3.geo.admin.chendpoints and overlapped in five core tools (layer search, identify, find, height, elevation profile). Maintaining two servers for one data source meant double the audits, double the CVE bumps, and an unclear choice for users. The rationale and the migration steps are documented in docs/merge-plan-swiss-geodata-mcp.md.This repository will be archived. The code below still works, but it receives no further fixes or dependency updates. See Migration below.
🇨🇭 Part of the Swiss Public Data MCP Portfolio
🗺️ swiss-geodata-mcp
CI PyPI Python License: MIT swiss-public-data-mcp
MCP server for Swiss federal geodata (geo.admin.ch).
Connects AI models to the federal geodata infrastructure: ~700 layers discoverable by keyword, spatial identify at any point, building zones (ARE), municipality lookup (swissBOUNDARIES3D), terrain heights and elevation profiles (swissALTI3D), and WGS84↔LV95 coordinate conversion. Part of the swiss-public-data-mcp portfolio. Private project, independent of any employer or institutional affiliation.
Demo query (anchor example)
In which building zone is the school building at Seilergraben 76 in Zurich,
and how high above sea level is it?
→ geo_zoning_at(2683531, 1247914) + geo_height(2683531, 1247914) returns the harmonised ARE zone and 411 m a.s.l. — verified live 2026-07-24.
Combined with swiss-housing-mcp:
What applies at this address?
→ address_to_egid("Seilergraben 76 Zürich") (housing) delivers the LV95 coordinates → geo_zoning_at + geo_municipality_at (geodata) deliver zone and municipality. If the GWR is the address book, geo.admin.ch is the atlas.
Tools (9)
| Tool | Description | Data source |
|---|---|---|
geo_search_layers |
Keyword search over the ~700-layer federal catalogue | geo.admin.ch SearchServer |
geo_identify |
What is at this LV95 point? (any layer) | geo.admin.ch MapServer identify |
geo_find |
Find features on a layer by attribute value | geo.admin.ch MapServer find |
geo_municipality_at |
Municipality + canton containing a point | swissBOUNDARIES3D |
geo_zoning_at |
Building zone(s) at a point | ch.are.bauzonen (ARE) |
geo_height |
Terrain height at a point | swissALTI3D height service |
geo_elevation_profile |
Elevation profile along a line | geo.admin.ch profile service |
geo_layer_info |
Queryable fields + legend (plain text) for a layer | geo.admin.ch MapServer |
geo_convert_coordinates |
WGS84 ↔ LV95 conversion | geodesy.geo.admin.ch reframe |
geo_search_layers is the discovery entry point that scales the whole catalogue without one tool per layer; geo_layer_info then reveals a layer's queryable fields for geo_find.
Tool annotations (MCP hints)
All tools are read-only (readOnlyHint: ✅, destructiveHint: ✗) and query live upstream services (openWorldHint: ✅). None are idempotent in the strict caching sense, as upstream data may change between calls.
Architecture decision
This server uses Architecture A (Live-API-only) — a deliberate deviation from the portfolio's dump-first default, documented per portfolio convention:
- The federal geodata infrastructure spans ~700 layers and terabytes; dump-caching is neither feasible nor useful.
api3.geo.admin.chis built exactly for point/feature queries and answered every probe reliably without authentication (live probe 2026-07-24: SearchServer, identify, find, height, profile, legend, layer metadata, reframe — all HTTP 200, No-Auth).- Consequence: no local cache, no TTL logic; every response carries
provenance: live_api.
Live probe findings (2026-07-24)
| Endpoint | HTTP | Status | Note |
|---|---|---|---|
SearchServer type=layers |
200 | ✅ works | catalogue full-text searchable |
| MapServer identify (bauzonen, boundaries) | 200 | ✅ works | tolerance 0 works for polygon layers |
| height service | 200 | ✅ works | value arrives as JSON string |
| profile.json (GET + geom) | 200 | ✅ works | COMB/DTM2/DTM25 altitudes |
MapServer {layer}/legend |
200 | ⚠️ HTML | stripped to plain text in geo_layer_info |
| reframe wgs84↔lv95 | 200 | ✅ works | coordinates arrive as JSON strings |
| Miss on identify/find | 200 | ⚠️ soft | empty results array — not an HTTP error |
Quick start
Claude Desktop
{
"mcpServers": {
"swiss-geodata": {
"command": "uvx",
"args": ["swiss-geodata-mcp"]
}
}
}
Cloud / Render.com (Streamable HTTP)
SWISS_GEODATA_TRANSPORT=streamable-http PORT=8000 swiss-geodata-mcp
Configuration
| Variable | Default | Purpose |
|---|---|---|
SWISS_GEODATA_TRANSPORT |
stdio |
stdio | streamable-http | sse |
HOST / PORT |
127.0.0.1 / 8000 |
HTTP binding (cloud transports only). Defaults to loopback; set HOST=0.0.0.0 explicitly to expose all interfaces in a cloud deployment. |
No API keys — Phase 1 is authentication-free.
Example queries
School planning
- «Which building zone applies at our planned school extension site?» →
geo_zoning_at - «Which municipality and canton is this coordinate in?» →
geo_municipality_at(bridges to BFS numbers used byswiss-statistics-mcpandswiss-housing-mcp) - «How steep is the school route between these two points?» →
geo_elevation_profile
Layer discovery
- «Is there federal data on noise exposure?» →
geo_search_layers("lärm")→geo_layer_info→geo_identify
See EXAMPLES.md for use cases grouped by audience (schools, parents, general public, developers) and a tool-selection reference table.
Testing
PYTHONPATH=src pytest tests/ -m "not live" # CI-safe (respx-mocked)
PYTHONPATH=src pytest tests/ -m live # against real upstream
Project structure
swiss-geodata-mcp/
├── src/swiss_geodata_mcp/
│ ├── server.py # FastMCP tools (9, prefix geo_*)
│ ├── geoadmin.py # geo.admin.ch client + retry + normalisation
│ ├── models.py # Pydantic v2 envelopes (source + provenance)
│ └── __main__.py # Dual-transport entry point
├── tests/ # respx-mocked + @pytest.mark.live
└── .github/workflows/ # CI + OIDC PyPI publish
Known limitations
- Registers live elsewhere: building/dwelling entities (EGID/EWID) belong to swiss-housing-mcp; this server is the spatial layer (zones, boundaries, heights). Deliberate separation to avoid twin servers.
- The harmonised zoning layer (ch.are.bauzonen) is an ARE synthesis; legally binding is only the cantonal/communal Nutzungsplanung (noted in every
geo_zoning_atresponse). - School-district polygons are municipal data (→
zurich-opendata-mcp), not federal; this server provides municipality boundaries, not Schulkreise. geo_identifyresult counts are capped upstream; area-wide aggregations are out of scope here (seebuildings_in_bboxin swiss-housing-mcp for the register case).- Coordinates must be LV95; WGS84 input fails fast with a pointer to
geo_convert_coordinates.
Changelog
See CHANGELOG.md
Contributing
Contributions are welcome — see CONTRIBUTING.md for how to report bugs, suggest a new layer, or submit code.
Security
This is a read-only, no-PII, public-open-data server. See SECURITY.md for the security posture and how to report a vulnerability.
License
MIT License — see LICENSE. Data: Swiss federal geodata infrastructure (geo.admin.ch / swisstopo and publishing federal offices), open government data with attribution.
Credits & related projects
- Data & services: geo.admin.ch (swisstopo), ARE, swisstopo geodesy
- Portfolio siblings: swiss-housing-mcp (register layer), swiss-statistics-mcp (statistics layer), zurich-opendata-mcp (municipal layer)
Author
malkreide · GitHub
Migration to swisstopo-mcp
Replace this server in your MCP client config:
// before
{ "swiss-geodata": { "command": "uvx", "args": ["swiss-geodata-mcp"] } }
// after
{ "swisstopo": { "command": "uvx", "args": ["swisstopo-mcp"] } }
Tool mapping
| this server | swisstopo-mcp |
note |
|---|---|---|
geo_search_layers |
swisstopo_search_layers |
— |
geo_identify |
swisstopo_identify_features |
— |
geo_find |
swisstopo_find_features |
— |
geo_height |
swisstopo_get_height |
— |
geo_elevation_profile |
swisstopo_elevation_profile |
takes a coordinate string; set coordinate_system="lv95" for LV95 support points |
geo_zoning_at |
swisstopo_zoning_at |
the non-binding ARE caveat now travels on every result record |
geo_municipality_at |
swisstopo_municipality_at |
BFS number is named bfs_commune_number and normalised to int |
geo_layer_info |
swisstopo_layer_info |
— |
geo_convert_coordinates |
swisstopo_convert_coordinates |
same direction values; same REFRAME service |
Two differences worth knowing
Coordinates. This server was LV95-only and rejected WGS84. swisstopo-mcp
accepts either lat/lon (WGS84) or easting/northing (LV95) on the
point-based tools — pass one pair, not both. Existing LV95 call sites keep
working; the argument names are the same.
Response shape. This server returned a JSON string (GeoEnvelope) with the
payload under result. swisstopo-mcp returns a structured ToolResponse:
records live in results (plural, always a list), with count, match_type,
source, license and a Markdown summary alongside. Code that parsed
result needs to read results.
Установка Swiss Geodata
У этого сервера нет опубликованного пакета — он собирается из исходников. Открой репозиторий и следуй инструкции в README.
▸ github.com/malkreide/swiss-geodata-mcpFAQ
Swiss Geodata MCP бесплатный?
Да, Swiss Geodata MCP бесплатный — установка в пару кликов через Unyly без оплаты.
Нужен ли API-ключ для Swiss Geodata?
Нет, Swiss Geodata работает без API-ключей и переменных окружения.
Swiss Geodata — hosted или self-hosted?
Доступен hosted-вариант: Unyly запускает сервер в облаке, локальная установка не обязательна.
Как установить Swiss Geodata в Claude Desktop, Claude Code или Cursor?
Открой Swiss Geodata на unyly.org, выбери вкладку своего клиента (Claude Desktop, Claude Code, Cursor) и нажми Install — конфиг сгенерируется автоматически, без правки JSON.
Похожие MCP
Fetch
Web content fetching and conversion for efficient LLM usage.
AWS KB Retrieval
Retrieval from AWS Knowledge Base using Bedrock Agent Runtime.
автор: modelcontextprotocolSpring AI MCP Server
Provides auto-configuration for setting up an MCP server in Spring Boot applications.
llm-analysis-assistant
A very streamlined mcp client that supports calling and monitoring stdio/sse/streamableHttp, and can also view request responses through the /logs page. It also
автор: xuzexin-hzMCP-Agent
A simple, composable framework to build agents using Model Context Protocol by [LastMile AI](https://www.lastmileai.dev)
автор: lastmile-aiSpring AI MCP Client
Provides auto-configuration for MCP client functionality in Spring Boot applications.
mcp.natoma.ai
A Hosted MCP Platform to discover, install, manage and deploy MCP servers by [Natoma Labs](https://www.natoma.ai)
MCPHub
Website to list high quality MCP servers and reviews by real users. Also provide online chatbot for popular LLM models with MCP server support.
MCP Servers Rating and User Reviews
Website to rate MCP servers, write authentic user reviews, and [search engine for agent & mcp](http://www.deepnlp.org/search/agent)
mkinf
An Open Source registry of hosted MCP Servers to accelerate AI agent workflows.
Compare Swiss Geodata with
Не уверен что выбрать?
Найди свой стек за 60 секунд
Автор?
Embed-бейдж для README
Похожее
Все в категории ai
