Activate when the user is WRITING APPLICATION CODE that programmatically interacts with a Kinetica GPU database — Python or Node apps, microservices, ETL scripts, or streaming data pipelines that import the gpudb SDK, call the Kinetica REST API from app code, or wire up Kinetica connectors (Spark, Kafka, JDBC, Airflow, Flink). Trigger on requests like "write a Python script that ingests this CSV into Kinetica", "build a FastAPI endpoint backed by Kinetica", "set up a Kafka to Kinetica pipeline", "use the gpudb client to upsert records", or any request for source code the user will save, version, and run as part of an application. Do NOT use for ad-hoc SQL, one-off table inspection, schema exploration, admin commands, or map tile rendering — those go to kinetica-execute.
Activate for INTERACTIVE EXPLORATION, QUERYING, and ADMINISTRATION of an existing Kinetica GPU database via its CLI — running SQL queries, inspecting table schemas and row counts, browsing data, one-shot data import/export, graph analytics (shortest path, PageRank, TSP), geospatial filtering, WMS map tile rendering (heatmaps, contours, rasters, class-breaks, labels, isochrones from x/y or WKT data), EXPLAIN plans, system table queries, table monitoring, security configuration, and database administration. Trigger on requests like "what tables are in schema X?", "how many rows in foo?", "show me the schema of bar", "find records where lat is in this bbox", "render a heatmap of column Y", "EXPLAIN this query", or any request where the user wants an answer FROM the database rather than code that talks to it. Activate even for simple tasks like checking table counts or running a quick query. Do NOT use when the user is writing an application, SDK client, or production data pipeline in source code — those go to ki
Activate when the user is WRITING APPLICATION CODE that programmatically interacts with a Kinetica GPU database — Python or Node apps, microservices, ETL scripts, or streaming data pipelines that import the gpudb SDK, call the Kinetica REST API from app code, or wire up Kinetica connectors (Spark, Kafka, JDBC, Airflow, Flink). Trigger on requests like "write a Python script that ingests this CSV into Kinetica", "build a FastAPI endpoint backed by Kinetica", "set up a Kafka to Kinetica pipeline", "use the gpudb client to upsert records", or any request for source code the user will save, version, and run as part of an application. Do NOT use for ad-hoc SQL, one-off table inspection, schema exploration, admin commands, or map tile rendering — those go to kinetica-execute.
Activate for INTERACTIVE EXPLORATION, QUERYING, and ADMINISTRATION of an existing Kinetica GPU database via its CLI — running SQL queries, inspecting table schemas and row counts, browsing data, one-shot data import/export, graph analytics (shortest path, PageRank, TSP), geospatial filtering, WMS map tile rendering (heatmaps, contours, rasters, class-breaks, labels, isochrones from x/y or WKT data), EXPLAIN plans, system table queries, table monitoring, security configuration, and database administration. Trigger on requests like "what tables are in schema X?", "how many rows in foo?", "show me the schema of bar", "find records where lat is in this bbox", "render a heatmap of column Y", "EXPLAIN this query", or any request where the user wants an answer FROM the database rather than code that talks to it. Activate even for simple tasks like checking table counts or running a quick query. Do NOT use when the user is writing an application, SDK client, or production data pipeline in source code — those go to ki
Bump the Kinetica plugin version across all 5 source files (2 SKILL.md frontmatter + 2 JSON configs + README.md version badge) and run build.sh to sync mirrored copies. Accepts patch (default), minor, or major.