GIS tools for AI agents: 65 free tools + 8 paid (hazard/site-scouting/GeoJSON export)
Scored across 112 tools
Many tools overlap heavily: assess_property_hazard, assess_property_hazard_x402, assess_site_due_diligence, assess_datacenter_site, assess_portfolio_climate_disclosure, and rank_hazard_safe_sites all wrap the same seismic/FEMA/NOAA hazard data. Bundles such as full_service_audit and audit_service also duplicate primitives like check_service_health, check_field_types, extract_domains, and find_layer_issues, forcing agents to guess which level of wrapper to use.
All names are snake_case, but several conventions coexist: verb-first commands (assess_*, get_*, check_*, bundle_*), noun_to_noun converters (geojson_to_csv, kml_to_shapefile), and object-first utilities (geometry_stats, centroids_geojson). Still readable, though not a single predictable pattern throughout the 112-tool surface.
112 tools is far beyond a well-scoped MCP surface and includes many paid/free variants, composite wrappers, and convenience bundles that could be consolidated. The count is extremely mismatched with the practical need for an agent to choose reliably.
The GIS surface is broad: geocoding, coordinate/CRS utilities, vector conversions, GeoJSON operations, ArcGIS service QA, hazard/site assessments, and map output are all covered. Gaps are relatively minor, such as limited raster analysis and no advanced network/isochrone tooling, but core GIS workflows are well represented.
Reverse-geocode a WGS84 point to country/region/locality via OSM Nominatim (no key, free — fair-use rate limited, best-effort). Returns JSON: {ok, display_name, country, country_code, region, locality, source}.
Inputs: lat (required), lon (required). Returns: result
PAID (100 credits) — nearby FAA-registered vertical obstructions (towers, stacks, wind turbines) at a US point, for telecom/tower-siting due diligence (e.g. "is there anything tall already near this proposed tower site", "does a 250ft tower here need FAA notification"). Source: official FAA Digital Obstacle File (652k+ records nationwide). If proposed_height_agl_ft is given, flags the general 14 CFR Part 77 200ft AGL notification threshold — a single well-documented rule, not a full aeronautical study (lower heights near airports can also trigger notification, not checked here). Requires Authorization: Bearer <api_key> and sufficient credits balance. Returns JSON: {ok, obstacle_count, obstacles, tallest_nearby_agl_ft, likely_requires_faa_notification, note}.
Inputs: lat (required), lon (required), proposed_height_agl_ft. Returns: result
PAID (300 credits) — data center site-scouting composite: grid proximity (power availability), multi-source hazard (seismic + FEMA + NOAA, required — same contract as assess_property_hazard), land cover, and a year of temperature history (cooling-load proxy). No invented overall suitability score — each source reported honestly. Requires Authorization: Bearer <api_key> and sufficient credits balance. Returns JSON: {ok, hazard, grid_proximity, land_cover, cooling_climate, note}.
Inputs: lat (required), lon (required). Returns: result
PAID (100 credits) — US Inflation Reduction Act "energy community" bonus tax credit qualification lookup for a renewable energy project site. Checks both DOE/NETL-published qualifying pathways (coal-mine- closure/coal-plant-retirement adjoining tracts, and MSA/non-MSA fossil- fuel-employment + above-average-unemployment areas) — a point qualifies if either matches. Reports qualification only, not a project-specific dollar estimate (the +10 percentage-point ITC/PTC bonus is the one publicly documented figure). Requires Authorization: Bearer <api_key> and sufficient credits balance. Returns JSON: {ok, lon, lat, qualifies_ for_bonus_credit, coal_closure_pathway, msa_fossil_employment_pathway, bonus_note, source}.
Inputs: lat (required), lon (required). Returns: result
One-call EV charging site-scouting indicator: grid proximity (power for fast-charging) + land cover + locality context, combined from GISGP's existing free primitives. No key needed. Returns JSON: {ok, grid_proximity, land_cover, location, physically_unsuitable}.
Inputs: lat (required), lon (required). Returns: result
Electric grid infrastructure proximity at a US point: nearby substations + transmission lines within 10 miles (community ArcGIS mirror of HIFLD data — proximity indicator, not a certified survey). No key needed. Returns JSON: {ok, substations: {count, nearest, source}, transmission_lines: {count, volt_classes, owners, source}}.
Inputs: lat (required), lon (required). Returns: result
Natural gas interstate/intrastate transmission pipeline proximity at a US point (EIA/HIFLD, 32,892 segments). Substitute for proprietary local-utility "call before you dig" data — covers major transmission pipelines only, NOT local distribution lines. Not a substitute for calling 811 before excavation. No key needed. Returns JSON: {ok, count, pipeline_types, operators, source, note}.
Inputs: lat (required), lon (required). Returns: result
PAID (200 credits PER PROPERTY, max 25 properties/call) — bulk physical climate-risk disclosure input for a real-estate/asset portfolio (SEC climate disclosure rule / EU CSRD physical-risk reporting). Runs assess_property_hazard for each [lon, lat] property and returns honest aggregate counts (properties with FEMA disaster history, properties with active NOAA alerts) alongside each property's own full result — no invented portfolio-level risk score. Use estimate_cost(tool_name= "assess_portfolio_climate_disclosure", units=) first to quote. Requires Authorization: Bearer <api_key>. Returns JSON: {ok, evaluated, scored_count, failed_count, failed, properties_with_disaster_ history_10y, properties_with_active_weather_alerts, properties, note}.
Inputs: properties (required). Returns: result
PAID (200 credits) — multi-source US property hazard indicator: USGS seismic risk (required), plus best-effort FEMA disaster declaration history (state, last 10 years) and NOAA real-time active severe weather alerts at the point. Requires Authorization: Bearer <api_key> and a sufficient credits balance — top up at https://gisgp.com/billing/mcp-credits/topup. Use estimate_cost first to quote without charging. Returns JSON: {ok, pga_percent_g, hazard_note, source, disaster_history_10y, active_weather_alerts}.
NOTE 2026-07-28: still not a full multi-hazard property report — no
flood-plain/wildfire-hazard polygon layer is reachable from GISGP's
infrastructure (FEMA NFHL/USFS geo-IP/WAF blocks), see hazard_scoring.py
module docstring and .claude/plans/kind-snacking-puppy.md. No invented
composite risk score — each source is reported honestly as what it is.
Inputs: lat (required), lon (required). Returns: result
PAID via x402 (USDC on Base Sepolia testnet, ~$0.20/call) — no GISGP account or API key needed, the agent pays directly on-chain. Multi-source US property hazard indicator (USGS seismic + FEMA disaster history + NOAA active alerts), same underlying data as assess_property_hazard. Returns JSON: {ok, pga_percent_g, hazard_note, source, disaster_history_10y, active_weather_alerts}.
Inputs: lat (required), lon (required).
US protected/conservation area proximity via USGS PAD-US (public domain, commercial-use OK — unlike Protected Planet/WDPA, whose API explicitly forbids commercial-entity use). Reports whether the point falls inside/near (2 mile radius) a federally-tracked protected area: name, managing agency, designation type, public access. US-only. Returns JSON: {ok, count, areas, source, note}.
Inputs: lat (required), lon (required). Returns: result
PAID (100 credits) — reforestation/afforestation site-scouting composite: terrain, land cover (flags already-forested/urban/water/snow as unsuitable), and a year of precipitation history (tree-viability proxy). All-free-primitive composite, priced for the bulk-screening convenience (carbon-offset/land-screening use case), no invented "sequestration score". Requires Authorization: Bearer <api_key> and sufficient credits balance. Returns JSON: {ok, terrain, land_cover, physically_unsuitable, precipitation, note}.
Inputs: lat (required), lon (required). Returns: result
One-call solar/wind site-scouting indicator: terrain + land cover + 1-year solar/wind climate averages + grid proximity, combined from GISGP's existing free primitives (no new external source). No key needed. Returns JSON: {ok, terrain, land_cover, solar_wind, grid_proximity}. Not a substitute for a professional feasibility study.
Inputs: lat (required), lon (required). Returns: result
PAID (300 credits) — master due-diligence report for a point: wraps assess_property_hazard (seismic + FEMA + NOAA, required) plus grid proximity, land cover, and reverse-geocoded locality — all in one call. No invented overall "site score" across sources. Requires Authorization: Bearer <api_key> and sufficient credits balance. Use estimate_cost first to quote. Returns JSON: {ok, hazard, grid_exposure, land_cover, locality, note}.
Inputs: lat (required), lon (required). Returns: result
PAID (400 credits) — checks whether two supposedly-redundant sites (e.g. two "backup" suppliers/warehouses/data centers) are actually independent: real geodesic distance between them + each site's own multi-source hazard profile (seismic + FEMA + NOAA) side by side, plus any FEMA disaster incident types they share in the last 10 years. No invented "redundancy score" — the caller judges independence from the real data. CAVEAT: FEMA disaster history is STATE-level, not point-level — two sites in the same state usually share most incident types regardless of true distance, so distance_km is the more reliable independence signal for same-state comparisons. Requires Authorization: Bearer <api_key> and sufficient credits balance. Returns JSON: {ok, distance_km, site_a, site_b, shared_fema_incident_types_10y, note}.
Inputs: site_a (required), site_b (required). Returns: result
One-call cell tower site-scouting indicator: grid proximity + terrain elevation (line-of-sight proxy) + land cover, combined from GISGP's existing free primitives. No key needed. Returns JSON: {ok, grid_proximity, terrain, land_cover, physically_unsuitable}.
Inputs: lat (required), lon (required). Returns: result
Real-time active wildfire proximity at a point via NASA FIRMS (VIIRS satellite hotspot detections, last 2 days) + electric grid exposure (transmission lines are a documented wildfire-ignition- liability factor). No key needed — GISGP holds its own free NASA FIRMS key server-side. Two separate honest indicators, no fabricated combined risk score. Returns JSON: {ok, active_fires, grid_proximity}.
Inputs: lat (required), lon (required). Returns: result
Full one-call QA report of an ArcGIS FeatureServer layer: health (latency, record count, capabilities) + field schema + coded domains + summary counts.
One call replaces separate health/fields/domains lookups. Returns JSON:
{name, geometry_type, health, fields, domains, summary}.
Inputs: service_url (required), token. Returns: result
Buffer every feature by a distance in METRES (negative shrinks).
Geodesically accurate at any latitude (each feature is buffered in a local
equidistant projection). Returns JSON {ok, distance_m, geojson} where geojson
is a FeatureCollection of buffer polygons.
Inputs: geojson (required), distance_m (required). Returns: result
PAID PER DELIVERY (30 credits each time it fires, NOT at creation — free to create/cancel) — the only MCP tool that exposes GISGP's core recurring-export product to agents: creates a schedule that re-exports a FeatureServer layer on its own and POSTs the file straight to your own webhook_url, no email/web UI account needed beyond the API key. Reuses the same scheduler that runs the paid web app's scheduled exports (fires within ~5 min of the due time).
format: "csv", "geojson", "shapefile", "kml", or "excel".
frequency: "hourly" (top of each hour), "daily" (at run_hour UTC), "weekly"
(at run_hour UTC on `weekday`, 0=Monday..6=Sunday), or "monthly" (at run_hour
UTC on `monthday`, 1-28).
webhook_url: must be a public, reachable HTTPS URL (validated at creation AND
at every delivery) — GISGP POSTs a JSON body {schedule_id, format, row_count,
filename, delivered_at, data_base64 (or download_url for files >5MB)}.
If the wallet lacks 30 credits when a delivery is due, that cycle is silently
skipped (schedule stays active, no error surfaced to you) — top up any time at
https://gisgp.com/billing/mcp-credits/topup?utm_source=mcp and the next cycle delivers normally.
Use estimate_cost or check_wallet_balance to plan ahead. Delete via
cancel_export_schedule when no longer needed — an abandoned schedule with an
empty wallet just skips forever, but does not charge or error.
Requires Authorization: Bearer <api_key>. Returns JSON: {ok, schedule_id,
next_run_at}.Inputs: format (required), service_url (required), webhook_url (required), token, +5 more. Returns: result
PAID (150 credits) — fetch ALL records from a live FeatureServer layer (no 50-record preview cap), auto-repair invalid geometries (GEOS make_valid), and scan for common problems, in one call. Replaces an uncapped query_features call (itself a paid-only capability) plus fix_geometry plus find_layer_issues. Requires Authorization: Bearer <api_key> and sufficient credits balance.
Returns JSON: {ok, feature_count, fixed_count, truncated, issues,
issue_count, geojson} — geojson is the fetched (and repaired) data.Inputs: service_url (required), token, where. Returns: result
PAID (120 credits) — fetch ALL records from a live FeatureServer layer once, then export the same data as Shapefile + CSV + GeoJSON + KML + Excel in a single ZIP (base64-encoded). Replaces an uncapped fetch plus 4 separate format-conversion calls. Requires Authorization: Bearer <api_key> and sufficient credits balance.
Returns JSON: {ok, feature_count, truncated, zip_base64, formats}.Inputs: service_url (required), token, where. Returns: result
PAID (30 credits PER ADDRESS, max 10 addresses/call) — batch geocode + enrich. For each address: forward-geocodes to lon/lat (free fallback chain: OSM Nominatim -> Photon -> US Census), then reverse-geocodes for country/region/locality context and classifies land cover at that point. Replaces 3 separate calls per address: geocode_address + analyze_location + classify_land_cover. Use estimate_cost(tool_name= "bundle_geocode_enrich", units=) first to quote. Requires Authorization: Bearer <api_key> and sufficient credits balance.
Returns JSON: {ok, results: [{address, ok, lon, lat, display_name,
geocode_source, country, region, locality, land_cover} or
{address, ok:false, error}], evaluated, failed_count}.Inputs: addresses (required). Returns: result
PAID (40 credits) — one-call file migration: convert a raw file to GeoJSON, then optionally reproject and simplify, then round coordinates. Replaces 2-4 separate free tool calls (convert + reproject + simplify + round) an agent would otherwise chain by hand — see estimate_cost's "replaces" field. Requires Authorization: Bearer <api_key> and sufficient credits balance.
input_format: "shapefile" (data = base64-encoded .zip), "kml", "gpx", or
"csv" (data = raw text for the latter three).
to_epsg: 0 skips reprojection (keeps the converter's native WGS84 output).
simplify_tolerance: 0 skips simplification (Douglas-Peucker, degrees).
round_decimals: coordinate rounding, always applied (default 6).
Returns JSON: {ok, feature_count, geojson, reprojection_notice}.Inputs: data (required), input_format (required), to_epsg, round_decimals, +1 more. Returns: result
PAID (100 credits) — migration-readiness report comparing two FeatureServer layers: schema diff, coded-value domains, and field types for both. Replaces compare_schemas + extract_domains×2 + check_field_types×2. Requires Authorization: Bearer <api_key> and sufficient credits balance.
Returns JSON: {ok, schema_diff, domains_a, domains_b, fields_a, fields_b,
ready, blockers} — ready is False and blockers lists the reasons if the
schema diff finds added/removed/changed fields or a geometry mismatch.Inputs: service_url_a (required), service_url_b (required), token_a, token_b. Returns: result
Real driving distance + duration between two points. Default: OSRM (public demo instance, no key — fair-use limited, not a production SLA). Optional agol_token: route through Esri's own World Route Service instead — uses YOUR OWN AGOL credentials/quota, not GISGP's (no extra charge either way, no GISGP account needed for this option). Returns "no_route" honestly when no driving path exists (e.g. across open water). Returns JSON: {ok, routable, distance_km, duration_min, source}.
Inputs: dest_lat (required), dest_lon (required), origin_lat (required), origin_lon (required), +1 more. Returns: result
FREE — cancel a schedule created by bundle_create_export_schedule. Only the API key that created it can cancel it. Returns JSON: {ok} or {ok: false, error}.
Inputs: schedule_id (required), token. Returns: result
Get the center point of each polygon/line feature — e.g. "find the center of each parcel", "convert polygons to points for labeling", "centroid of each feature". Properties are kept.
Returns JSON {ok, geojson} — a FeatureCollection of points.
Inputs: geojson (required). Returns: result
Inspect the field schema of an ArcGIS FeatureServer layer.
Returns JSON: {layer_name, geometry_type, fields:[{name, alias, type,
length, nullable, has_domain}]} — useful before importing/mapping data.
Inputs: service_url (required), token. Returns: result
Check whether an ArcGIS FeatureServer layer is reachable and responsive.
Returns JSON: {ok, ms (latency), name, geometry_type, field_count,
record_count, capabilities, max_record_count}.
Inputs: service_url (required), token. Returns: result
FREE — check your own MCP credits wallet balance (the same account paid tools like assess_property_hazard charge against). Requires Authorization: Bearer <api_key> — an agent can call this on its own, no browser/dashboard needed. Returns JSON: {ok, balance, eur_equivalent} or {ok: false, error} if no API key was provided.
Inputs: no named parameters. Returns: result
Land cover class at a WGS84 point via ESA CCI Land Cover 2018 (ArcGIS Living Atlas ImageServer, no key, 300m resolution). Returns JSON: {ok, class_code, class_label, source}.
NOTE 2026-07-26: gated separately (mcp_layer1_landcover, Off pending live
smoke-check) — swapped from the earlier ESA WorldCover WMS source, which
was confirmed dead (being phased out). This ArcGIS REST identify endpoint
was live-verified locally (Sofia -> class 8, "Artificial Surface or
Urban Area") before this swap.
Inputs: lat (required), lon (required). Returns: result
Compare the field schemas of two ArcGIS FeatureServer layers.
Returns JSON diff: {added, removed, changed:[{field, diffs}], same,
geometry_match, identical} — useful for QA before an import or migration.
Inputs: service_url_a (required), service_url_b (required), token_a, token_b. Returns: result
Convert coordinate pairs between EPSG coordinate systems (max 1000 points).
points: [[x, y], ...] in from_epsg axis order (lon/lat for EPSG:4326).
Returns JSON: {"points": [[x, y], ...], "from_epsg": ..., "to_epsg": ...}.
Inputs: points (required), to_epsg (required), from_epsg (required). Returns: result
GISGP Lite (a 24-hour pass, a 7-day pass or a monthly plan) — pass the api_key of that
account (or Authorization: Bearer). convert one GIS
file between formats with the full site converter: input shapefile (.zip), KML/KMZ, GPX, CSV,
GeoJSON, GPKG, FlatGeobuf, MapInfo TAB (.zip, source="mapinfo-tab"), FileGDB (.zip,
source="filegdb"), DXF or DWG; to geojson | shapefile | kml | kmz | gpx | csv | xlsx | gpkg |
flatgeobuf | mapinfo-tab | wkt | dxf. source_epsg for projected CSV / a shapefile without .prj,
target_epsg to reproject (GeoJSON-composable pairs). File up to 4.5 MB inline (base64). Returns
JSON {ok, filename, content_type, size, data_base64} or {ok, download_url} for big results,
{ok: false, error, pairs} otherwise.
Inputs: to (required), filename (required), data_base64 (required), source, +3 more. Returns: result
Draw the smallest polygon that wraps around all input features (a convex hull) — e.g. "draw a boundary around these points", "find the outer perimeter of my data", "convex hull of these locations".
Returns JSON {ok, geojson} — a single polygon.
Inputs: geojson (required). Returns: result
Count features in an ArcGIS FeatureServer layer, optionally filtered by a WHERE clause.
service_url: full layer URL ending in /FeatureServer/<n>. token: only for private layers.
Returns JSON: {"ok": true, "count": ..., "name": ..., "geometry": ...}.
Inputs: service_url (required), token, where. Returns: result
Convert a CSV/spreadsheet with lat/lon (or X/Y) columns to GeoJSON — auto-detects the coordinate columns (e.g. "map this spreadsheet", "CSV with lat/lon to GeoJSON", "turn my Excel export into map points"). Returns a GeoJSON FeatureCollection as a JSON string.
Inputs: csv (required). Returns: result
Convert a CSV with latitude/longitude columns directly to GPX waypoints (2026-09-03) — e.g. "these coordinates to a GPX for my GPS". Column names are auto-detected (lat/latitude/y, lon/lng/longitude/x).
Inputs: csv (required). Returns: result
Detect whether [x, y] pairs are [lon, lat] or [lat, lon] — resolves a genuinely common real-world confusion (one of the most-viewed GIS StackExchange questions ever, 550k+ views). Certain when any value exceeds ±90 (only longitude can); honestly reports 'inconclusive' otherwise, never a fabricated guess. Pure computation, no external API. Returns JSON: {ok, order, confidence, note}.
Inputs: points (required). Returns: result
Keyword search across ArcGIS Hub Open Data — real public FeatureServer/MapServer datasets from government/organization ArcGIS Online portals worldwide. Returns real service_url values usable directly with query_features/count_features/rest_explore/compare_schemas. Free, no API key. Not a universal open-data search (ArcGIS Hub coverage only — see tool note). Returns JSON: {ok, query, count, results, source, note}.
Inputs: query (required), limit. Returns: result
PAID (30 credits) — open-data discovery with the metadata an agent needs BEFORE querying: up to 50 ArcGIS Hub results (free tool: 10) and, for the top 25, each layer's geometry type, field list, record count and query capability — so you can pick the right layer in one call instead of probing each service_url. Same coverage as the free discover_open_geodata (ArcGIS Hub Open Data), no key, real REST metadata. Needs an API key with credits — use estimate_cost first.
Returns JSON: {ok, query, count, enriched, results:[{name, type,
source_org, description, service_url, layer_name, geometry_type, fields,
field_count, record_count, max_record_count, supports_query}]}.
Inputs: query (required), limit, enrich. Returns: result
Merge overlapping/adjacent geometries into one (optionally grouped by a
property field by → one dissolved feature per distinct value).
Returns JSON {ok, by, group_count, geojson}.
Inputs: geojson (required), by. Returns: result
Convert an AutoCAD DXF file's contents to GeoJSON — e.g. "convert
this DXF to GeoJSON", "AutoCAD drawing to GIS format", "DXF export from
Civil 3D to GeoJSON", "extract only the UTILITIES layer from this DXF".
POINT/LINE/LWPOLYLINE/POLYLINE/ARC/CIRCLE/ELLIPSE/SPLINE/HATCH/MPOLYGON convert
(curves tessellated into LineString/Polygon), TEXT/MTEXT become points
with a text property, and block INSERTs are expanded at their real
position/scale/rotation with ATTRIB values as properties; 3D solids and
meshes have no 2D GIS equivalent and are skipped, not errored on.
Every feature's properties.layer carries its DXF layer name (group code
8). Pass layer to return only that layer's entities — see
list_dxf_layers to discover what layer names exist first.
Inputs: dxf (required), layer. Returns: result
PAID (150 credits) — render a DXF drawing to a one-page PDF site-plan report (vector plot + entity-type breakdown + layer list, base64- encoded PDF) — e.g. "make a PDF of this site plan", "DXF to PDF report". No basemap (DXF coordinates are usually local drawing units, not real lon/lat) — use georeference_dxf first if you need a real-world-referenced report. Requires Authorization: Bearer <api_key> and sufficient credits balance. Returns JSON: {ok, pdf_base64, feature_count}.
Inputs: dxf (required), title. Returns: result
Ready-to-paste embed code for a live interactive map — thin wrapper around share_map (same live page, same 30-day TTL), for embedding directly in a website/dashboard rather than sharing a link. Returns JSON: {ok, embed_html, url, expires_at}.
Inputs: geojson (required), name, width, height. Returns: result
Get the bounding box (min/max rectangle) around each feature — e.g. "get the bounding box of this shape", "bbox for each feature", "rectangular extent of my geometry". Properties are kept.
Returns JSON {ok, geojson}.
Inputs: geojson (required). Returns: result
Suggest the correct UTM EPSG code to project a WGS84 lon/lat into for accurate metric measurements (buffer/area/distance).
Returns JSON {ok, epsg, utm_zone, hemisphere, name, warning}.
Inputs: lat (required), lon (required). Returns: result
FREE — quote a paid tool's cost WITHOUT executing it or charging the
wallet (call this before assess_property_hazard etc.).
units = candidate/item count for per-unit-priced tools (currently only
rank_hazard_safe_sites — pass the number of candidates you intend to
submit); ignored (treated as 1) for every flat-priced tool. Returns JSON:
{ok, tool_name, units, credits, eur_estimate} for credits-wallet tools;
for assess_property_hazard_x402 (paid on-chain via x402, not the credits
wallet) returns {ok, tool_name, units, payment, price_usd} instead; or
{ok: false, error} for an unknown/free tool name.
Inputs: tool_name (required), units. Returns: result
PAID (80 credits) — convert a GeoJSON FeatureCollection to an Excel workbook (.xlsx, base64-encoded). Point geometries get LATITUDE/LONGITUDE columns; other geometry types get a GEOMETRY_WKT column. Separate from the web app's own Excel export (AGOL-service-tied, quota-gated there). Requires Authorization: Bearer <api_key> and sufficient credits balance. Returns JSON: {ok, xlsx_base64, feature_count}.
Inputs: geojson (required). Returns: result
PAID (150 credits) — render an ad-hoc GeoJSON FeatureCollection to a one-page PDF report (static map + property table, base64-encoded PDF). Separate from the web app's own PDF Builder (AGOL-service-tied, quota- gated there) — this is for GeoJSON an agent already has in hand. Requires Authorization: Bearer <api_key> and sufficient credits balance. Returns JSON: {ok, pdf_base64, feature_count, truncated, size_bytes}.
Inputs: geojson (required), title, accent. Returns: result
Extract coded value domains from an ArcGIS FeatureServer layer.
Returns JSON: {"layer_name": ..., "domains": {field: {name, field_alias, codes}}}.
Inputs: service_url (required), token. Returns: result
Scan an ArcGIS FeatureServer layer for common problems: missing ObjectID, disabled query, stale data, all-null fields, features without geometry.
Sample-based (first 200 records). Returns JSON: {layer_name,
issues:[{severity, code, message}], issue_count, sample_size, note}.
Inputs: service_url (required), token. Returns: result
Repair invalid geometries (self-intersections, bad rings) via GEOS make_valid. Already-valid geometries are left untouched. Returns JSON {ok, fixed_count, feature_count, geojson}.
Inputs: geojson (required). Returns: result
Convert a FlatGeobuf file (base64-encoded .fgb) to a GeoJSON FeatureCollection. FlatGeobuf is a cloud-native binary format with a built-in spatial index, used for fast HTTP-range-request streaming (e.g. deck.gl, MapLibre, QGIS cloud layers).
Returns the GeoJSON FeatureCollection as a JSON string.
Inputs: flatgeobuf_base64 (required). Returns: result
Express one WGS84 point in every common notation at once: decimal degrees, DMS, DDM, UTM (zone/band/easting/northing), MGRS and USNG, British National Grid reference (inside GB only), Geohash, Plus Code, Web Mercator metres, the ArcGIS Online "lon, lat" form, the best local UTM EPSG code, plus ready-to-open Google Maps / Street View / Google Earth / Apple / Bing / OpenStreetMap / Waze links. Pure computation. mgrs_precision: 1-5 digit pairs (10 km ... 1 m), default 5. Returns JSON: {ok, decimal_degrees, dms, ddm, utm, mgrs, usng, bng, geohash, plus_code, web_mercator, agol_format, local_utm_epsg, links}.
Inputs: lat (required), lon (required), mgrs_precision. Returns: result
Complete one-call QA report: A-F grade + full audit (health/fields/ domains/summary) + issue scan, merged into one response — replaces 3 separate calls (grade_service + audit_service + find_layer_issues).
Returns JSON: {ok, grade, score, categories, archetype, insight,
percentile, share_url, audit: {name, health, fields, domains, summary},
issues, issue_count}.
Inputs: service_url (required), token. Returns: result
GISGP Lite (a 24-hour pass, a 7-day pass or a monthly plan) — pass the api_key of that
account (or Authorization: Bearer). contour lines for a box bbox=[west, south, east, north] (degrees) or a
GeoJSON polygon, from Copernicus 30 m (source="cop30", worldwide, up to 100 km² per box) or
USGS 10 m ("usgs10", US, up to 25 km²). interval in units ("m" or "ft"; omit for an
automatic interval), format geojson | kml | kmz | shp | dxf | dxf3d | landxml | preview,
crs an EPSG code for the DXF / LandXML (default: the UTM zone). Returns JSON: GeoJSON
with stats for geojson/preview; for files {ok, filename, data_base64} or {ok, download_url}.
Inputs: crs, bbox, units, format, +5 more. Returns: result
Render a GeoJSON FeatureCollection/Feature to a static PNG map image (base64 data URL) — for embedding in reports/emails/documents, unlike share_map's interactive live page. basemap: 'topo', 'streets', or 'satellite'. Reuses the exact rendering path already used inside PDF map reports, no new map-rendering logic. Optional agol_token: for a single-Point input, renders through Esri's own Static Maps Service instead (real ArcGIS basemap tiles) — uses YOUR OWN AGOL credentials/quota, not GISGP's (multi-feature/non-point input always uses the default renderer regardless of token). Returns JSON: {ok, image_data_url, feature_count, width, height, basemap}.
Inputs: geojson (required), width, accent, height, +2 more. Returns: result
Forward-geocode a free-text address to lon/lat. Default: OSM Nominatim (no key, free, fair-use rate limited) — the complement to analyze_location's reverse geocode. Optional agol_token: route through Esri's own World Geocoding Service instead for more authoritative matches — uses YOUR OWN AGOL credentials/quota, not GISGP's (no extra charge either way, no GISGP account needed for this option). Returns JSON: {ok, lon, lat, display_name, type, importance, source}.
Inputs: address (required), agol_token. Returns: result
Compare two GeoJSON FeatureCollections — added/removed/changed features.
Matched by an `id_field` property if given, else by array position (exact
structural comparison, not topological equals). Returns JSON {ok, added,
removed, changed, added_count, removed_count, changed_count, unchanged_count}.
Inputs: geojson_a (required), geojson_b (required), id_field. Returns: result
Convert GeoJSON to CSV/spreadsheet — one row per feature, properties as columns, geometry flattened into a column (e.g. "turn this GeoJSON into Excel/CSV", "I need a spreadsheet from these map features", "export GeoJSON as a table"). Returns the CSV as plain text.
Inputs: geojson (required). Returns: result
Convert GeoJSON to an AutoCAD DXF file — e.g. "export this GeoJSON to DXF for AutoCAD", "GIS data to CAD format", "convert to .dxf for Civil 3D". Points become DXF POINT entities; lines and polygon rings become LWPOLYLINE entities. A feature's properties.layer (if present) is written back as the DXF entity's layer (group code 8), defaulting to layer "0" otherwise.
Inputs: geojson (required). Returns: result
Convert GeoJSON to FlatGeobuf (.fgb) — the cloud-native binary format with a built-in spatial index. FlatGeobuf holds one geometry type per file, so mixed point/line/polygon input is automatically split into one .fgb per category and returned inside a zip.
Returns JSON: {flatgeobuf_base64|flatgeobuf_zip_base64: str, split: bool,
split_categories: [str]|null}.
Inputs: geojson (required). Returns: result
Convert GeoJSON to a GPX file — e.g. "export this as a GPS track", "GeoJSON to GPX for my Garmin", "make waypoints from this GeoJSON". Points become GPX waypoints; lines and polygon rings become GPX tracks (GPX has no native polygon type). Non-name properties are dropped — GPX has no generic attribute-table equivalent.
Inputs: geojson (required). Returns: result
Convert GeoJSON to KML for Google Earth / Google My Maps (e.g. "open this in Google Earth", "GeoJSON to KML", "make this viewable in Google My Maps"). Returns the KML as plain text.
Inputs: geojson (required). Returns: result
Convert GeoJSON to a MapInfo TAB file set, ready for MapInfo Pro. TAB is a multi-sidecar-file format (.tab/.dat/.id/.map) and single geometry type per file, like Shapefile — always returned as a zip, mixed point/line/polygon input is split into one .tab set per category.
Returns JSON: {mapinfo_tab_zip_base64: str, split: bool,
split_categories: [str]|null}.
Inputs: geojson (required). Returns: result
Convert GeoJSON to an Esri Shapefile ZIP — the format ArcGIS Desktop Pro/QGIS/most legacy GIS software still expects (e.g. "I need a shapefile for ArcGIS Pro", "GeoJSON to SHP", "convert for QGIS import").
Returns JSON: {shapefile_zip_base64, warnings}.
Inputs: geojson (required). Returns: result
Convert a GeoJSON geometry (or a Feature's geometry) to WKT — the format needed to insert geometry into PostGIS/SQL (e.g. "I need this for a PostGIS INSERT", "GeoJSON to WKT", "geometry as WKT string").
Inputs: geojson (required). Returns: result
Complete one-call GeoJSON health check: validate + auto-repair invalid geometry + stats (area/length/vertices) + bbox envelope, merged — replaces 4 separate calls (validate_geojson + fix_geometry + geometry_stats + envelope_geojson).
Returns JSON: {ok, valid, errors, warnings, fixed_count, stats, envelope,
geojson} where geojson is the (possibly auto-fixed) input.
Inputs: geojson (required). Returns: result
Compute area (m²/km²), length (m/km), vertex count, centroid and bbox of GeoJSON.
Metric measurements use an equal-area projection so area/length are real metres.
Inputs: geojson (required). Returns: result
Reproject a DXF drawn in a local/arbitrary CAD coordinate system into real-world coordinates — e.g. "this DXF isn't in real coordinates, fix it", "georeference this site plan", "I know 2 points on this drawing's real location, convert the rest". Give exactly 2 control points as a JSON array: '[{"local":[x,y],"real":[lon,lat]}, {"local":[x,y],"real":[lon,lat]}]' — a 2-point similarity transform (uniform scale + rotation + translation) is solved and applied to every coordinate. Returns a GeoJSON FeatureCollection in the real coordinate system implied by the 2 points.
Inputs: dxf (required), control_points (required). Returns: result
Daily historical weather for a point/date range via Open-Meteo's ERA5-based archive (no key, free, data back to 1940). Dates are YYYY-MM-DD. Returns JSON: {ok, daily: {time, ...}, source}.
Inputs: lat (required), lon (required), end_date (required), start_date (required), +1 more. Returns: result
GISGP Lite (a 24-hour pass, a 7-day pass or a monthly plan) — pass the api_key of that account (or Authorization: Bearer). height above mean sea level of one point from Copernicus GLO-30 (30 m), read from the public elevation model (GLO-90 fallback). Returns JSON {ok, elevation_m, source, resolution_m, credit}; elevation_m is null at sea. For a free, coarser answer use get_terrain_profile.
Inputs: lat (required), lon (required), api_key. Returns: result
PAID (50 credits) — elevation PROFILE along a line, not a single point: pass points=[[lon,lat],…] (up to 100) or start=[lon,lat] + end=[lon,lat] with samples (2-100, evenly spaced). Returns every point's elevation and distance plus min/max, total gain/loss, net change, mean and max segment slope. For "how steep is this route", "line-of-sight terrain between two towers", "elevation change along this pipeline/trail". Copernicus DEM GLO-90 via Open-Meteo (~90 m). The free get_terrain_profile gives one point; this gives the whole line + the maths. Needs an API key with credits — use estimate_cost first.
Returns JSON: {ok, points, distance_m, min/max/start/end_elevation_m,
total_gain_m, total_loss_m, net_change_m, mean_slope_pct,
max_segment_slope_pct, profile:[{i, lon, lat, distance_m, elevation_m}]}.
Inputs: end, start, points, samples. Returns: result
Nearest USGS real-time stream/river gauge to a point (US rivers and streams only): live streamflow (ft3/s) and gauge height (ft), reported honestly as provisional USGS data — no derived flood-risk score, since USGS does not publish per-site flood-stage thresholds through this API. First real-time (not static/historical) primitive in the catalog. No key needed. Returns JSON: {ok, found, site_name, site_no, distance_miles, streamflow_cfs, gauge_height_ft, observation_time, source, note}.
Inputs: lat (required), lon (required). Returns: result
Get the ground elevation (metres above sea level) at a coordinate — e.g. "what's the elevation here", "is this point in a flood-risk lowland", "elevation at this address". Via Open-Meteo (Copernicus DEM GLO-90, no key, free, ~90m resolution).
Returns JSON: {ok, lon, lat, elevation_m, source}.
Inputs: lat (required), lon (required). Returns: result
Convert a GPS track file (GPX) directly to CSV (2026-09-03) — waypoints get LATITUDE/LONGITUDE columns, tracks/routes get a GEOMETRY_WKT column instead.
Inputs: gpx (required). Returns: result
Convert a GPS track/route file (GPX — from Garmin, Strava exports, hiking/cycling apps, waypoints) to GeoJSON (e.g. "convert my GPS track to GeoJSON", "GPX to GeoJSON", "import my Strava route").
Returns the GeoJSON FeatureCollection as a JSON string.
Inputs: gpx (required). Returns: result
Convert a GPS track file (GPX) directly to KML for Google Earth — e.g. "open my Garmin/Strava track in Google Earth", "GPX to KML".
Inputs: gpx (required). Returns: result
Convert a GPS track file (GPX) directly to an Esri Shapefile ZIP — waypoints become points, tracks/routes become lines (2026-09-03). Returns JSON: {shapefile_zip_base64, warnings}.
Inputs: gpx (required). Returns: result
Grade an ArcGIS FeatureServer layer A–F across 5 categories (schema quality, data completeness, performance, maintenance, configuration).
Returns JSON: {grade, score, categories:[{key, label, score, note}], archetype,
insight, percentile, share_url} — share_url is a public scorecard page you can
send to a human. token: only for private layers (never stored in the scorecard).
Inputs: service_url (required), token. Returns: result
GISGP Lite (a 24-hour pass, a 7-day pass or a monthly plan) — pass the api_key of that
account (or Authorization: Bearer). which coordinate
reference system puts these easting/northing rows where they belong. lines are text rows
such as "P-7 690183.66, 4729859.76" (up to 1,000); say roughly where with country ("Kenya")
or lat/lon. Returns JSON {ok, confidence: high|medium|low|none, candidates: [{epsg, name,
area, swapped, national, km, lat, lon}], rows, bad_rows}. Then convert with the chosen EPSG via
reproject_geojson or /api/v1/crs/apply.
Inputs: lines (required), lat, lon, api_key, +1 more. Returns: result
Best-guess candidate CRS for raw [x, y] coordinates with no metadata (e.g. a shapefile with a missing .prj) — real demand signal (79k+ views on GIS StackExchange). Genuinely ambiguous by nature — returns ranked candidates with an honest 'not a certain identification' caveat, never a single confident answer. Pure computation, no external API. Returns JSON: {ok, x_range, y_range, candidates, note}.
Inputs: points (required). Returns: result
Convert Google Earth / Google My Maps KML (2.0–2.2) to GeoJSON — the common request when someone exported a map from Google Earth or Google My Maps and needs it in a format other GIS tools/APIs accept (e.g. "convert my Google Earth file to GeoJSON", "KML to GeoJSON").
Returns the GeoJSON FeatureCollection as a JSON string.
Inputs: kml (required). Returns: result
Convert a Google Earth/My Maps KML file directly to GPX — waypoints from Points, tracks from Lines/Polygons — e.g. "KML to GPX for my Garmin", "load this map on my GPS" (2026-09-03, completes the reverse of the existing gpx_to_kml).
Inputs: kml (required). Returns: result
Convert a Google Earth/My Maps KML file directly to an Esri Shapefile ZIP — e.g. "Google Earth file to shapefile for ArcGIS", "KML to SHP". Returns JSON: {shapefile_zip_base64, warnings}.
Inputs: kml (required). Returns: result
List the distinct DXF layer names used in a drawing — e.g. "what
layers does this DXF have", "list layers before I extract one". Real
CAD exports organize entities into named layers (BOUNDARY, UTILITIES,
TEXT, ...); use the result with dxf_to_geojson's layer param to pull
out just one. Returns a JSON array of layer name strings.
Inputs: dxf (required). Returns: result
Convert a MapInfo TAB file set (base64-encoded .zip containing .tab + its .dat/.map/.id sidecar files) to a GeoJSON FeatureCollection. MapInfo TAB is the native format of MapInfo Pro — a multi-file format on disk, always uploaded as a zip.
Returns the GeoJSON FeatureCollection as a JSON string.
Inputs: mapinfo_tab_zip_base64 (required). Returns: result
Smallest-area rectangle wrapping all input features combined — can be tilted to fit tighter than a straight bounding box (e.g. "tightest rectangle around this data", "minimum bounding rectangle", "oriented bounding box"). Returns JSON {ok, geojson}.
Inputs: geojson (required). Returns: result
For each feature in A, find the nearest feature in B by centroid-to-
centroid geodesic distance. Attaches _nearest properties and
_nearest_distance_m. Returns JSON {ok, geojson}.
Inputs: geojson_a (required), geojson_b (required). Returns: result
Move a WGS84 point by a geodesic distance (metres) along a compass bearing (0=N, 90=E, 180=S, 270=W). Pure computation, no external API. Returns JSON: {ok, lon, lat, origin_lon, origin_lat, distance_m, bearing_deg, source}.
Inputs: lat (required), lon (required), distance_m (required), bearing_deg (required). Returns: result
Combine two GeoJSON FeatureCollections with a boolean set operation.
op: intersection | difference (a minus b) | symmetric_difference | union.
Each side is unioned into one geometry first, then the operation is
applied. Returns JSON {ok, op, geojson} — empty geojson means no overlap,
not an error.
Inputs: geojson_a (required), geojson_b (required), op. Returns: result
Parse ANY single coordinate string a human might paste into WGS84
lat/lon — decimal degrees with any separator or an EU decimal comma,
DMS (42°41'51.7"N 23°19'18.9"E), DDM / GPS style (N 42° 41.535'),
NMEA (4241.535,N,02319.314,E), UTM with a zone/band (34T 583000 4727000),
MGRS (34TFN8300027000), British grid refs (TQ 30 80), Geohash, full Plus
Codes, and Google/Apple/Bing/OSM map links. Detects lon,lat order from
hemisphere letters, labels or out-of-range values and says so in note.
Deterministic, no LLM — LLMs are unreliable past 2-3 decimals; this is
exact. Ambiguous input (e.g. "583000 4727000" with no zone) is an error
with a concrete hint, never a guess.
Returns JSON: {ok, kind, label, lat, lon, note, ...notation fields}.
Inputs: text (required). Returns: result
GISGP Lite (a 24-hour pass, a 7-day pass or a monthly plan) — pass the api_key of that
account (or Authorization: Bearer). a legal description
→ where it is: q like "T2N R3E Sec 14", "NW¼ SE¼ Sec 12 T3N R4W", "Township 1 North, Range 2
East" or BLM-coded "1460N 0920W SEC 8"; state the two-letter code (tried in every PLSS state
when empty). Returns JSON {ok, results: [{label, state, lat, lon, bbox, plssid, geometry…}],
read_as, notes}.
Inputs: q (required), state, api_key. Returns: result
GISGP Lite (a 24-hour pass, a 7-day pass or a monthly plan) — pass the api_key of that account (or Authorization: Bearer). US Public Land Survey System township, range, section and meridian at a point (BLM CadNSDI). Returns JSON {ok, found, township, range, section, meridian, plssid, …} or {ok, found: false, note}.
Inputs: lat (required), lon (required), api_key. Returns: result
Fetch actual feature records (attributes + geometry) from a FeatureServer layer.
Free preview capped at 50 records (WHERE filter + chosen fields). Returns JSON:
{returned, total, truncated, geojson, note}. Full unlimited export is a paid operation.
Inputs: service_url (required), limit, token, where, +1 more. Returns: result
Server-side aggregate stats on a numeric field of a FeatureServer layer.
No records fetched — AGOL computes sum/avg/min/max/count (also stddev/var).
Optional WHERE filter and group_by field. Returns JSON: {field, results:[...]}.
Cheap way to answer "what's the average/total of field X" without an export.
Inputs: field (required), service_url (required), stats, token, +2 more. Returns: result
PAID (200 credits PER CANDIDATE, max 25 candidates/call) — bulk seismic- risk screener. Runs assess_property_hazard's seismic check under the hood for each [lon, lat] candidate and returns only the top_n ranked by lowest real seismic risk (PGA). Not a full multi-hazard ranking — ranked by ONE real, single-source metric only (matches this project's no-invented- composite-score stance); disaster_history_10y/active_weather_alerts are still included per-candidate for context. Use estimate_cost(tool_name= "rank_hazard_safe_sites", units=) first to quote. Requires Authorization: Bearer <api_key>. Returns JSON: {ok, evaluated, excluded_count, excluded, ranked_by, top, note}.
Inputs: candidates (required), top_n. Returns: result
Bulk solar/wind site screener — evaluates up to 25 candidate points (each {"lon": ..., "lat": ...}) and returns only the top_n, ranked by average daily solar radiation. Candidates with unsuitable land cover (urban/water/snow) or no substation nearby are filtered out, not silently scored low. No invented composite score. No key needed. Returns JSON: {ok, evaluated, excluded_count, excluded, ranked_by, top}.
Inputs: candidates (required), top_n. Returns: result
Round every coordinate to N decimal places to shrink a GeoJSON file's size — e.g. "my GeoJSON file is too big", "reduce coordinate precision", "strip excess decimal places". 6 decimals ≈ 11cm precision, plenty for most mapping use cases.
Inputs: geojson (required), decimals. Returns: result
Change a GeoJSON's coordinate reference system (CRS/projection) — e.g. "reproject to Web Mercator", "convert WGS84 to UTM", "my coordinates are in the wrong CRS", "EPSG:4326 to EPSG:3857". Reprojects every geometry type between any two EPSG codes.
Inputs: geojson (required), to_epsg (required), from_epsg (required). Returns: result
List the layers and tables of an ArcGIS FeatureServer/MapServer root.
Point at a service root to enumerate its layers/tables, or at a single
layer URL to describe it. Returns JSON: {is_server, layers:[{id, name, type}]}.
Inputs: service_url (required), token. Returns: result
Convert a Shapefile ZIP (base64-encoded .zip with .shp/.dbf/.shx[/.prj]) to a GeoJSON FeatureCollection. Reprojects to WGS84 if a .prj says otherwise.
Returns JSON: {geojson: {...}, reprojection_notice: str|null}.
Inputs: shapefile_zip_base64 (required). Returns: result
Convert a Shapefile ZIP (base64-encoded .zip with .shp/.dbf/.shx[/.prj]) directly to GPX — points become waypoints, lines/polygons become tracks (2026-09-03). Returns GPX 1.1 XML as plain text.
Inputs: shapefile_zip_base64 (required). Returns: result
Reduce the number of vertices in GeoJSON geometry (Douglas-Peucker algorithm) — e.g. "simplify this polygon", "this shape has too many points", "reduce vertex count for faster rendering". Tolerance in degrees; lower = more detail kept.
Inputs: geojson (required), tolerance. Returns: result
Attach properties from every B feature that satisfies a spatial
predicate to each A feature (as a _matches list).
predicate: intersects | within | contains | touches | crosses | overlaps.
Returns JSON {ok, predicate, match_count, geojson}.
Inputs: geojson_a (required), geojson_b (required), predicate. Returns: result
Convert a WGS84 lon/lat + zoom level to Slippy Map (XYZ) tile x/y, that tile's lon/lat bbox, and its Bing quadkey.
Returns JSON {ok, zoom, x, y, tile_bbox, quadkey}.
Inputs: lat (required), lon (required), zoom (required). Returns: result
Check whether a GeoJSON file/string is valid — catches broken geometry, malformed structure, self-intersections, and out-of-range coordinates before you try to use the file elsewhere (e.g. "is this GeoJSON valid", "why won't my GeoJSON load", "lint my GeoJSON", "check topology errors"). Validates RFC 7946 structure, topology, and WGS84 coordinate ranges.
Returns a JSON report: {valid, errors, warnings, stats}.
Inputs: geojson (required). Returns: result
Divide space into regions closest to each input point (Voronoi diagram) — e.g. "find the service area for each location", "which points are closest to which store", "Voronoi/Thiessen polygons". Non-point features use their centroid. Needs at least 2 points. Returns JSON {ok, cell_count, geojson}.
Inputs: geojson (required). Returns: result
Convert a WKT (Well-Known Text) geometry string — the format PostGIS/SQL geometry columns print by default — to GeoJSON (e.g. "this came from a PostGIS query", "WKT to GeoJSON", "parse this POLYGON(...) string").
Inputs: wkt (required). Returns: result
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Geocoding, routing, isochrones, traffic, weather, and place search for AI agents. 19 MCP tools.
Japan POI tools for AI agents: 2.94M facilities, station catchments, chains, grants and signals
62 real-world tools for agents: search, scraping, social, enrichment, image, video, voice.
Geopolitical grounding for AI agents: country risk, forecasts, chokepoints, sanctions. Free tier.
Add one secure layer between your agents and this server.
