pyside6-qml-models-services
Pass
Audited by Gen Agent Trust Hub on Feb 27, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The FileDiscoveryService in services/file_discovery.py contains a potential indirect prompt injection surface where a malicious identifier could lead to directory traversal.
- Ingestion points: The job_number parameter in the discover_files method is used to construct a directory path.
- Boundary markers: No boundary markers or instructions are present to ensure that the agent ignores directory traversal sequences (e.g., '..') provided in the input.
- Capability inventory: The service has the capability to crawl and index the file system using pathlib's rglob and stat methods.
- Sanitization: The job_number is joined to the base_path without validation or normalization, allowing access to files outside the intended directory scope.
Audit Metadata