Mode d’emploi
- Copiez la commande et examinez le paquet avant de l’installer dans votre environnement agent.
- Consultez le document original pour les outils, identifiants et environnements requis.
- Décrivez les entrées et le livrable souhaité ; comparez la sortie aux sources.
npx skills add https://github.com/cartodb/agent-skills --skill carto-connect-datawarehouseAvant de commencer
- L’installation n’inclut ni abonnement au modèle, ni quota API, ni accès aux fichiers.
- Étoiles et installations reflètent l’activité, pas un test indépendant de qualité ou de sécurité.
Document Skill original
Rendu depuis le dépôt source en conservant titres, exemples, code, tableaux, liens et images.
carto-connect-datawarehouse
CARTO runs spatial analytics in the user's own data warehouse. A connection is the bridge between CARTO and that warehouse: it carries credentials, target project/database scoping, and sometimes a service account or PAT. Most other CARTO operations (querying, importing, building maps, running workflows) require an existing connection.
| Task | MCP | CLI (fallback) |
|---|---|---|
| List / inspect connections | explore_data (list_connections on any session; get_connection is OAuth-only) | carto connections list / get |
| Create / update a connection | manage_connections (create, update) — OAuth session only | carto connections create / update |
| Delete a connection | delete (kind=connection) — OAuth session only | carto connections delete |
Access-path routing. With the MCP server attached, list overexplore_data(list_connections); create/update overmanage_connections; delete overdelete. On a token session onlylist_connectionsis available —get_connectionand the write tools (manage_connections,delete) need OAuth, so fall back to thecarto connectionsCLI. Also use the CLI when the server isn't attached or for scripted setups. The engine-choice guidance and pitfalls below apply on either path. Detection signals: `carto-basics/references/access-paths.md`.
When to use this skill
- The user wants to connect a new warehouse to CARTO.
- The user is debugging a connection (auth failures, missing tables, permission errors).
- A downstream skill needs a connection name and you don't yet know which engine the user has.
- The user is rotating credentials or moving from one project/database to another.
Use `carto-explore-datawarehouse` once a connection exists and you want to inspect what's inside it.
Quick lifecycle (CLI)
bash
carto connections list --json # what's already connected? (--all, --search "prod")
carto connections get <id> # detailed view of one connection
carto connections create # interactive create
carto connections update <id> # rotate credentials, change scoping
carto connections delete <id> # remove (irreversible)List/inspect are non-destructive — run them freely before deciding what to do. If the user already has a connection, don't push a new one — use the existing one.
Choosing an engine
| Engine | When to choose it | Reference |
|---|---|---|
| BigQuery | Google Cloud users; CARTO's flagship integration; rich GIS functions native. | references/bigquery.md |
| Snowflake | Snowflake-shop customers; geospatial via SQL functions and native types. | references/snowflake.md |
| Redshift | AWS-shop customers on Redshift Serverless or RA3 clusters. | references/redshift.md |
| Postgres | Self-hosted or RDS Postgres with PostGIS; common for small/medium deployments. | references/postgres.md |
| Databricks | Lakehouse / Unity Catalog users; SQL Warehouses for interactive workloads. | references/databricks.md |
| Oracle | Oracle Database with Spatial; on-prem or OCI / Autonomous Database. | references/oracle.md |
Common pitfalls
- Auth-mode mismatch: BigQuery supports OAuth (interactive) and service-account JSON (CI). Pick one consistently per environment; mixing the two breaks shared connections.
- Region vs project: Some engines need both an account/project and a region (Snowflake, Redshift). Skipping the region typically yields "endpoint not found" rather than a permissions error.
- Default database/schema scoping: CARTO writes tilesets, named sources, and analytics output back into the warehouse. Confirm which dataset/schema CARTO may write to before creating the connection.
- Permissions for describe and table reads come from the credential CARTO holds, not the user's CARTO role. A CARTO Admin with a low-privilege service account still sees "permission denied" from the warehouse.
What this skill doesn't cover
- Browsing tables/schemas of an existing connection — `carto-explore-datawarehouse`.
- Running SQL against the warehouse — `carto-query-datawarehouse`.
- Importing files into the warehouse — `carto-import-export-data`.
