developing-country.org

Use this when the user asks what is at, near or around a place in Sri Lanka (a town, address, listing or map point). Given the place's lat/lng (and optional radius_km, default 10), returns a provenance-tagged readout: district demographics (point-in-polygon), nearest expressway + status, nearby planned developments, attractions, waterfalls, trails, cycling routes, parks, and your own prospects. Completeness is first-class: known_absent means we have no geometry (silent zones, MoH clinics, courts) — do not invent polygons; osm_only is not an official register. Gazette 924/12 defines a 100 m silent-zone facility class; we do not ship those buffers as law. Use query_data for bounded spatial or indicator-table research; this tool stays a point readout. Each run is logged privately to the signed-in account history (`saved.url` map link). If the user gives a listing URL: extract location, then call with lat/lng plus source_url and title.

analyze_location

Find and understand Sri Lanka datasets. No arguments: a slim index of all Indicator Tables (with themes, publisher, time span, geography) and spatial entities. {q} and/or {theme}/{publisher}: ranked search across titles, descriptions, field names and dimension values (e.g. q:"remittances", q:"Jaffna", theme:"ECON"). {dataset_id}: one dataset card — exact field ids with roles/units, every filterable value, time span, geography join key, caveats, provenance and freshness, ready-to-run preset calls with chart presentations, and ranked related datasets with reasons. {full:true}: the complete provenance catalog (large; same as resource developing-country://catalog). Every response has a `next` list of suggested calls.

get_layer_catalog

List all prospects owned by the authenticated user, newest first.

list_prospects

Use this to read statistics, rankings, trends or map features from a dataset found with get_layer_catalog, and to save them as a live chart or map. Bounded queries over every registered platform Dataset. op=select reads one Indicator Table: dataset_id plus exact field ids, optional where/order_by/limit (rows capped at 200, or 5000 when saved with a presentation), or aggregate it with group_by (dimension fields) + aggregates [{fn: sum|mean|min|max|count, field, as}] instead of fields (sum only on fields the table declares summable). Get the field ids, filterable values and ready-to-run preset calls from get_layer_catalog {dataset_id} first — never guess a field id. Spatial ops (bbox, nearest, point_in_polygon, overlay, gap, choropleth_join, isochrone) take an ontology entity in `entities`; bbox with include_geometry (+ where property filters) returns any GeoJSON entity's own geometry, e.g. polygons, for a saved map; large shapes are not inlined in the text (geometry_omitted, geometry_ref names the asset and selection) while every feature row is. Without presentation the result is a preview the MCP App tables and charts inline; add presentation to persist a live, citable view_url (requires a signed-in verified account). Chart: family, dimension, measure, optional series and highlight. Map: optional camera, annotations [{id, coordinates:[lng,lat], text}], and layers merged onto the result layers by id (entity ids for spatial ops; the dataset_id for a select) with label, style {geometry fill|line|circle, color token or #hex, colorBy {field, scale sequential|diverging|threshold|categorical, breaks, colors, labels, classLabels}, radius, radiusBy, labelField}, legend and join. A select becomes a map when every row has numeric lng/lat or its layer has join {tileset:"demographics", sourceLayer: gnd|dsd|district|province, rowKey}. A saved view also returns `publication` {dataView, datasetVersions, sourceManifest}: the exact CMS evidence_bundle entries (developing-country://evidence/{view_id} re-reads it). Every result carries provenance, completeness and a `next` list of ready-to-send follow-up calls.

query_data

Register a property prospect. Provide a listing url (the server fetches og:title/description/image and any lat/lng), and/or explicit title + lat/lng. Private by default. Returns the stored prospect with scraped metadata plus `web_url` — a map link you can give the user (it appears live under "My prospects" in the web app).

add_prospect

Use this to find a named place in Sri Lanka and its coordinates (e.g. before analyze_location). Searches by name across searchable ontology stores (metro stops and lines, expressways, developments, BOI zones, attractions, waterfalls, trails, cycling, parks) plus your prospects. Does not search the metro catchment, sea-route or administrative-boundary layers; use query_data for those. Returns matches with layer, coordinates, and provenance.

search_places

Update a prospect you own: title, notes, verdict, visibility (private|link|public), or coordinates.

update_prospect