Convert coordinates between China map systems (BD09/GCJ02/WGS84/Mercator) and calculate geodesic distance and polygon area via @zealgeo/mcp-geo-server

@zealgeo/mcp-geo-server v1.2.0 — verified recipe

Install: npm install @zealgeo/mcp-geo-server Entry: dist/cli.js (stdio MCP server) Deps: @modelcontextprotocol/sdk, zod (18.1 kB unpacked) Credential-free: yes (pure local computation, no network calls)

Tools (3)

Conversion methods (8)

BD09toGCJ02, GCJ02toBD09, WGS84toGCJ02, GCJ02toWGS84, BD09toWGS84, WGS84toBD09, WebMercatortoLngLat, LngLattoWebMercator

Distance units: meters, kilometers

Area units: square_meters, square_kilometers, mu (Chinese land unit), hectares

coordType: WGS84 (default), GCJ02, BD09

Key gotchas

1. Tool descriptions are in Chinese (Mandarin) — all descriptions, property labels, and error messages are zh-CN. The tool names and JSON structure are English.
2. Method enum values are CamelCase — WGS84toGCJ02 NOT wgs84_to_gcj02. Wrong case returns Zod validation error with correct options list.
3. Unit enums use underscores — square_meters NOT squareMeters, square_kilometers NOT squareKilometers.
4. Outside China, WGS84↔GCJ02 returns identity — coords pass through unchanged (correct behavior: GCJ02 offset only applies within China's boundaries).
5. Uses Web Mercator projection for distance/area — NOT geodesic/Haversine. At Central Park latitude (~40.8°N), area is ~34% smaller than real geodesic area. Distance errors are smaller but still projection-dependent. Suitable for small to medium scale calculations, not globe-spanning.
6. Polygon auto-closes — no need to repeat first coordinate as last.
7. First call ~158ms (JIT), subsequent 1-12ms.

Verified trace (15 calls, 100% success, p50=4ms)

# 1. WGS84→GCJ02 (Beijing Forbidden City)
call: mcp_geo_convert({method:"WGS84toGCJ02", longitude:116.397128, latitude:39.916527})
→ {output: {longitude:116.40337, latitude:39.91743}} [158ms]
  Offset: +0.006° lon, +0.001° lat (~550m shift)

# 2. GCJ02→WGS84 (reverse)
call: mcp_geo_convert({method:"GCJ02toWGS84", longitude:116.403694, latitude:39.921204})
→ {output: {longitude:116.39745, latitude:39.92030}} [21ms]

# 3. WGS84→BD09 (GPS → Baidu)
call: mcp_geo_convert({method:"WGS84toBD09", longitude:116.397128, latitude:39.916527})
→ {output: {longitude:116.40974, latitude:39.92377}} [3ms]
  BD09 adds secondary offset on top of GCJ02

# 4. BD09→WGS84
call: mcp_geo_convert({method:"BD09toWGS84", longitude:116.410154, latitude:39.923036})
→ {output: {longitude:116.39754, latitude:39.91579}} [5ms]

# 5. WGS84 → Web Mercator (EPSG:3857)
call: mcp_geo_convert({method:"LngLattoWebMercator", longitude:116.397128, latitude:39.916527})
→ {output: {longitude:12957269.02, latitude:4853819.61}} [4ms]

# 6. Web Mercator → WGS84
call: mcp_geo_convert({method:"WebMercatortoLngLat", longitude:12958095.89, latitude:4852834.05})
→ {output: {longitude:116.40456, latitude:39.90974}} [4ms]

# 7-8. BD09↔GCJ02 round-trip — both OK [1ms, 12ms]

# 9. Distance Beijing→Shanghai
call: mcp_geo_calculate_distance({coordinates:[[116.397,39.917],[121.474,31.230]], unit:"kilometers"})
→ {distance: 1318.37 km} [5ms]

# 10. Polyline Istanbul→Ankara→Izmir
call: mcp_geo_calculate_distance({coordinates:[[28.978,41.008],[32.860,39.933],[27.143,38.424]], unit:"meters"})
→ {distance: 1132123 m} [3ms]

# 11. Distance with GCJ02 coordType — converts to WGS84 internally
call: mcp_geo_calculate_distance({coordinates:[[116.404,39.921],[121.474,31.230]], unit:"kilometers", coordType:"GCJ02"})
→ {distance: 1318.52 km} [5ms]

# 12. Area Central Park (rectangle)
call: mcp_geo_calculate_area({coordinates:[[-73.96