The Agent Skills Directory

[COMMAND_EXECUTION]: Several scripts in the examples folder utilize subprocess.run() to orchestrate external GIS and build tools.
examples/established-stack/analysis.py executes dbt to build SQL models.
examples/established-stack/delivery.py executes npm ci and npm run build to compile Observable Framework views, and openmapstack.integrations.qgis for desktop project generation.
examples/tartu-development/routing.py invokes valhalla_build_tiles to generate routing graphs.
examples/tartu-development/qgis_delivery.py uses docker run to execute PyQGIS headlessly.
These executions are integral to the GIS workflow and operate on project-local resources.
[EXTERNAL_DOWNLOADS]: The skill's examples and instructions involve downloading GIS data and software dependencies from well-known and trusted sources.
Fetches data from official Estonian registries (Maa- ja Ruumiamet S3, ETAK WFS) and Tartu City Government GIS.
Downloads OSM snapshots from Geofabrik (download.geofabrik.de).
Uses standard package registries (NPM via unpkg.com and PyPI).
References the Protomaps basemap service provided by the vendor (goplex.ee).
[DYNAMIC_EXECUTION]: The toolkit uses templating engines to generate code and data views.
examples/established-stack/analysis.py uses Jinja2 to render SQL model files.
examples/tartu-development/dashboard-template.html uses string replacement to embed derived data into the final dashboard.
[INDIRECT_PROMPT_INJECTION]: The skill defines a surface for processing untrusted external data (GeoJSON, CSV, OSM PBF).
Ingestion points are clearly defined in the pipeline.py templates and examples (e.g., read_parquet, ST_Read).
The instructions emphasize validation, schema checks, and the use of structured data (JSON) to mitigate risks associated with processing external content.

open-map-stack