What Is Mapular MCP?
Understand what Mapular MCP is, what the three servers do, and how the CLI ties them together.
Mapular MCP lets any MCP-compatible AI assistant — Claude Code, Claude Desktop, Cursor — query location data, run spatial SQL, and build interactive map dashboards using natural language.
It ships as two MCP servers and a CLI.
The three pieces
mapular-mpoi
A read-only data server. Queries pre-indexed points of interest from Mapular's location database across 10+ countries. Data is spatially indexed by H3 cells and served as Parquet files for downstream analysis.
Typical use: "Download all pharmacies in Munich" → Parquet file on disk.
mapular-geo
A spatial SQL engine built on DuckDB with spatial and H3 extensions. Takes Parquet files from mapular-mpoi and lets you run aggregations, proximity queries, geocoding, and exports.
Typical use: "Find all pharmacies within 500 m of this address" → SQL result.
mapular-admin
A write server for managing mapbooks. Creates and configures interactive map dashboards, adds data layers, and generates share links.
Typical use: "Create a mapbook showing pharmacy density across Berlin" → share link.
The CLI
The mapular CLI handles authentication and wires the servers into your AI client.
mapular login # authenticate once via browser
mapular install # write server config into .mcp.json and Claude Desktop
After mapular install, the servers start automatically whenever you open Claude Code from that directory.
How the servers work together
A typical end-to-end session:
mpoi_list_regions— discover available countriesmpoi_fetch— download filtered POI data as a Parquet filequery(mapular-geo) — run spatial SQL on the filecreate_mapbook+add_series(mapular-admin) — turn results into a map dashboardshare_mapbook— generate a link to share with stakeholders