qgis
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides Python scripts and CLI recipes that ingest and process external spatial data files (Shapefiles, GeoPackages) and database connections (PostGIS).
- Ingestion points: Input parameters in
run_geoprocessing(SKILL.md),apply_categorized_symbology(gemini.md), and theINPUTparameter in theNormalizeAttributesAlgorithmclass (gpt.md). - Boundary markers: Absent. The scripts do not implement specific delimiters or instructions to ignore embedded commands within the spatial data metadata or attributes.
- Capability inventory: The skill utilizes
qgis.corefor file writing viaQgsVectorFileWriterandQgsFeatureSink, and suggests executing system commands via theqgis_processCLI. - Sanitization: Absent. The skill assumes input data adheres to expected GIS formats without performing content sanitization for potential prompt injection strings within attribute fields.
Audit Metadata