qgis-impl-postgis

Pass

Audited by Gen Agent Trust Hub on Jun 18, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is educational in nature, providing robust documentation on how to use QGIS core classes (QgsDataSourceUri, QgsAuthManager) safely.
  • [DATA_EXPOSURE]: The skill includes dedicated sections (Critical Warnings and Anti-Patterns) that explain how to prevent credential leaks. It emphasizes using QgsAuthManager for encrypted storage and the uri(False) method to prevent sensitive data from appearing in logs or metadata files.
  • [PROMPT_INJECTION]: The skill facilitates the ingestion of data from external PostGIS databases. While it correctly handles connection security, there is an inherent risk of indirect prompt injection if an agent processes untrusted text content retrieved from a database table.
  • Ingestion points: Data is brought into the agent context via QgsVectorLayer and conn.executeSql (referenced in SKILL.md and references/examples.md).
  • Boundary markers: The skill encourages the use of QgsDataSourceUri for safe URI building but does not provide specific instructions for sanitizing the actual data content returned by queries.
  • Capability inventory: The skill enables SQL execution, database schema discovery, and file export operations.
  • Sanitization: No explicit sanitization of database content is demonstrated; users should ensure that data retrieved from external databases is treated as untrusted before being used in further prompts.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 18, 2026, 12:10 AM
Security Audit — agent-trust-hub — qgis-impl-postgis