swiss-geodata-mcp
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
<!-- mcp-name: io.github.malkreide/swiss-geodata-mcp -->
π¨π Part of the Swiss Public Data MCP Portfolio
πΊοΈ swiss-geodata-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 WGS84βLV95 coordinate conversion. Part of the swiss-public-data-mcp portfolio. Private project, independent of any employer or institutional affiliation.
π©πͺ Deutsche Version
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?
β geo_zoning_at(2683531, 1247914) + geo_height(2683531, 1247914) returns the harmonised ARE zone and 411 m a.s.l. β verified live 2026-07-24.
Combined with swiss-housing-mcp:
What applies at this address?
β address_to_egid("Seilergraben 76 ZΓΌrich") (housing) delivers the LV95 coordinates β 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 β 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: β
, destructiveHint: β) and query live upstream services (openWorldHint: β
). 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) β 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 β 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 | β works | catalogue full-text searchable |
| MapServer identify (bauzonen, boundaries) | 200 | β works | tolerance 0 works for polygon layers |
| height service | 200 | β works | value arrives as JSON string |
| profile.json (GET + geom) | 200 | β works | COMB/DTM2/DTM25 altitudes |
MapServer {layer}/legend |
200 | β οΈ HTML | stripped to plain text in geo_layer_info |
| reframe wgs84βlv95 | 200 | β works | coordinates arrive as JSON strings |
| Miss on identify/find | 200 | β οΈ soft | empty results array β 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 β Phase 1 is authentication-free.
Example queries
School planning
- Β«Which building zone applies at our planned school extension site?Β» β
geo_zoning_at - Β«Which municipality and canton is this coordinate in?Β» β
geo_municipality_at(bridges to BFS numbers used byswiss-statistics-mcpandswiss-housing-mcp) - Β«How steep is the school route between these two points?Β» β
geo_elevation_profile
Layer discovery
- Β«Is there federal data on noise exposure?Β» β
geo_search_layers("lΓ€rm")βgeo_layer_infoβgeo_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/
βββ src/swiss_geodata_mcp/
β βββ server.py # FastMCP tools (9, prefix geo_*)
β βββ geoadmin.py # geo.admin.ch client + retry + normalisation
β βββ models.py # Pydantic v2 envelopes (source + provenance)
β βββ __main__.py # Dual-transport entry point
βββ tests/ # respx-mocked + @pytest.mark.live
βββ .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 (β
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 β 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 β 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 Β· GitHub
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.