raleigh
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches civic datasets, transit feeds, and documentation from official City of Raleigh domains (raleighnc.gov, goraleigh.org), Wake County government sites, and established service providers like ArcGIS and Azure US Gov Cloud. These are legitimate sources for the skill's purpose.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes data from a wide variety of external APIs and HTML sources. It implements strong defensive measures to prevent malicious data from influencing the agent or user, such as stripping terminal control characters from remote text and neutralizing spreadsheet formulas in CSV exports to prevent formula injection.\n
- Ingestion points: Data enters the skill context through City of Raleigh Drupal nodes, RSS feeds, ArcGIS Hub collections, and the RWECC active incident feed.\n
- Boundary markers: Data export functions in arcgis.py and geocode.py use explicit quoting for potential formula triggers.\n
- Capability inventory: Restricted to read-only HTTPS requests (with one acknowledged HTTP exception for fire reports) and atomic file writes.\n
- Sanitization: content is cleaned of control characters in the public_safety_stats and rfd_reports modules before processing.\n- [REMOTE_CODE_EXECUTION]: No remote code execution or dynamic code generation patterns were detected. All dependencies are standard libraries or packages from trusted registries.
Audit Metadata