Skills catalog
The CARTO Agent Skills catalog ships 23 skills organized in three tiers. The list below is a snapshot; the canonical, always-up-to-date catalog is skills/catalog.json in the public repo.
Utility tier
Foundational CARTO primitives. No dependencies; loaded by other skills as a shared base.
Start here for first-time CARTO use. Orients on the two access paths — the MCP Server (the primary path for inline, in-conversation work) and the CLI (for scripting, CI/CD, headless and bulk/cross-org work, and deeper administration) — and covers CLI install, authentication, profiles, JSON output, and async job patterns.
Choose and configure the data warehouse engine connection for CARTO (BigQuery, Snowflake, Redshift, Postgres, Databricks, Oracle).
Write spatial SQL against the connected warehouse — dialect-specific guidance, performance defaults, and CARTO's query/job execution model.
Discover what's in the connected warehouse — schemas, tables, columns, and CARTO named sources.
Platform tier
CARTO product surfaces. Depend only on utility skills.
Import geospatial files into the data warehouse via CARTO, export results back out, and prepare tilesets for fast map rendering.
Build, schedule, and operate analytics DAGs in CARTO Workflows — the no-code/low-code orchestration layer over the data warehouse. Triggers when the user wants to author a workflow, run/edit one, schedule a DAG, or copy a workflow across profiles or orgs.
Discover and subscribe to external spatial datasets via CARTO Data Observatory and partner catalogs.
Administer the CARTO org — users, roles, quotas, activity audit, and bulk resource operations.
Author, edit, publish, and validate CARTO Builder maps. Routes to the MCP Server's map tools (create_map, update_map, validate_map, read_maps, view_map) when attached, and to the carto maps CLI for scripted/bulk authoring, cross-org copy, screenshots, or when the server isn't attached.
Render an ad-hoc interactive map inline in the chat from a deck.gl declarative spec via the CARTO MCP Server's view_map tool. For exploratory / throwaway visualizations; triggers on "show me X on a map", "visualize Y", "make a heatmap of Z".
Preview an existing saved CARTO Builder map inline in the chat via the CARTO MCP Server — resolve the map with read_maps (by name or search), then render it with view_map by map ID. Renders a lightweight read-only preview. Triggers on "show me the X map", "open the Y map", and inline previews of a freshly-created map.
Generate a working geospatial application powered by CARTO and deck.gl: basemap, layers (vector / H3 / quadbin / raster), widgets, filters, legend, inputs, optional chat-with-map agent, and the right auth strategy (public token, OAuth, SSO, or M2M).
Use-case tier
Recipe skills that compose platform skills into end-to-end spatial analyses. Each carries trigger keywords so the agent routes on user intent.
Builds Getis-Ord Gi* hotspot analysis workflows. Triggers on: hotspots, coldspots, spatial clusters, Getis-Ord, Gi*, cluster detection, concentration areas, spacetime hotspot, temporal clusters, time-varying patterns, hotspot trends, emerging hotspots, Mann-Kendall.
Builds Moran's I spatial autocorrelation workflows. Triggers on: spatial autocorrelation, Moran's I, spatial dependency, HH/HL/LH/LL quadrants, LISA, spatial weight matrix, classifying locations into cluster types.
Builds Geographically Weighted Regression (GWR) workflows. Triggers on: GWR, geographically weighted regression, spatially varying relationships, local regression, local coefficients, spatial non-stationarity.
Spatial enrichment workflows. Triggers on: enrich, add demographics, estimate population around locations, compute spatial features, sociodemographic analysis, trade-area enrichment.
Trade area and catchment analysis. Triggers on: trade area, catchment area, isochrone, drive time, walk time, billboard placement, OOH audience targeting, scoring candidate locations.
Site selection and cannibalization analysis. Triggers on: site selection, cannibalization, new store location, optimal location, twin areas, look-alike areas.
Territory planning combining territory balancing and location allocation. Triggers on: sales territories, service zones, balanced territories, location allocation, depot/hub placement, response-time optimization, demand coverage.
Routing and origin-destination analysis. Triggers on: routing, travel time, OD matrix, isoline, isochrone, catchment area, drive-time polygon, accessibility analysis, commute patterns, OD flow, shortest-path computation.
Geocoding workflows that convert addresses or place names into coordinates. Triggers when the user has tabular data with addresses but no geometry column.
Composite-score workflows. Triggers on: composite scores, indexes, multi-variable rankings, market potential, resilience indexes, risk indexes, weighted scores, PCA, supervised/unsupervised scoring.
End-to-end ArcGIS Portal / AGOL to CARTO migration. Three phases over a shared MIGRATION_MANIFEST.md: discover (enumerate and classify items), migrate-data (Hosted Feature Layers / Tables → CARTO tables), migrate-maps (Web Maps and simple Dashboard / Web Experience / Web Mapping Application entries → CARTO Builder maps). Default mode is batch; migrated maps are tagged From ArcGIS and created as private. Triggers when the user wants to migrate ArcGIS content to CARTO, asks "what's in my Portal", or names a Portal/AGOL URL.
Dependencies and routing
Use-case skills depend on carto-create-workflow. Platform skills depend on the relevant utility skills (e.g. carto-create-workflow depends on carto-basics, carto-connect-datawarehouse, and carto-query-datawarehouse; carto-render-inline-map and carto-preview-builder-map depend only on carto-basics). When a use-case skill is triggered, the agent loads it along with its dependency chain.
Routing happens via skill descriptions and trigger keywords (the catalog descriptions above). Phrasing your request with one of the trigger keywords is the most reliable way to route into a specific skill.
Last updated
Was this helpful?