Case study · 2026
Wi-Fi Explorer for Wireless@SGX
A Singapore connectivity guide for digital nomads, remote workers and curious wanderers: find listed venues, read the floor details, open walking directions.
1,306mapped venues
1,800source records
5MCP tools
Built on the openly licensed data.gov.sg catalogue. The map and list work in the browser; location is optional and asked for only on demand; map tiles load only after you choose Show map.
The problem
Hotspot lists are scattered, and “nearest Wi-Fi” questions are exactly what people ask an assistant. But an honest answer needs to say what the data can and can’t prove.
What I built
- Map and search: responsive venue map plus a searchable list that works without map tiles.
- Read-only JSON API shared by every integration.
- MCP server over stdio or stateless Streamable HTTP with five tools:
nearest_hotspots,search_venues,search_hotspots,get_hotspot,dataset_info. - Automation demos: an importable n8n workflow and an Activepieces custom piece prototype.
- Quality: locked dependencies, Docker/Compose, provenance checks, and tests that use SDK MCP clients over stdio and HTTP (run locally).
Design decisions
- It’s a source-backed guide, not a live monitor. Listings don’t prove public access, signal, hours, seating or power.
- Unknown origins fail loudly instead of inventing coordinates.
- The API flags incomplete rankings (
ranking_complete:false) so clients can disclose them. - Distances are straight-line metres, not walking distance or radio range.
Independent project, not affiliated with IMDA or the operators. Public edition data: data.gov.sg catalogue (IMDA), Singapore Open Data Licence; dataset page dated 6 June 2024, each record dated 18 March 2020. Figures shown are for this public edition. Listings are historical, not a current inventory.
Stack
Node 22, MCP SDK, OpenStreetMap tiles, n8n, Activepieces, Docker.
