This skill should be used when the user asks to "inspect", "describe", "preview", "analyze", or "look inside" a GIS or CAD file (`.shp`, `.gdb`, `.dwg`, `.dxf`, `.geojson`, `.kml`), or asks "what's in this shapefile / geodatabase / DWG", "show me the layers", "what CRS does this use", "how many features", "list the attributes". Invoked as `/gis-to-db:inspect <file> [--json] [--features-sample N]`. Read-only — never writes to the user's filesystem outside `/tmp` and never touches a database.
Installation
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
This skill should be used when the user asks to "inspect", "describe", "preview", "analyze", or "look inside" a GIS or CAD file (`.shp`, `.gdb`, `.dwg`, `.dxf`, `.geojson`, `.kml`), or asks "what's in this shapefile / geodatabase / DWG", "show me the layers", "what CRS does this use", "how many features", "list the attributes". Invoked as `/gis-to-db:inspect <file> [--json] [--features-sample N]`. Read-only — never writes to the user's filesystem outside `/tmp` and never touches a database.
This skill produces a structured analysis of a GIS or CAD file — its format, geometry type(s), layers, feature count, CRS, attribute schema, and a small sample of features — without writing anything to a database. It is the safe first step before any ingestion.
When to Use
Invoke when the user wants to understand a file's contents before deciding how to ingest it. Common phrasings:
Validator agent runs first. Invoke the gis-preflight-validator agent via the Task tool with mode=inspect to confirm the file exists, is readable, and GDAL is available (host or Docker).
Decide execution context. Prefer the host's Python+GDAL if available; otherwise run inside the configured Docker image (docker_image from settings, default osgeo/gdal:ubuntu-small-3.8.0).
Run the inspection script. The skill emits scripts/gis_inspect.py to /tmp/gis_inspect.py (or mounts it into Docker), then executes it with the file path and flags.
Render the report. Parse the script's JSON output and render the markdown table (or pass through JSON if --json).