learning-research-designer
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [Prompt Injection] (SAFE): No instructions attempting to bypass safety filters or override agent behavior were found in the metadata or descriptions.- [Data Exposure & Exfiltration] (SAFE): No hardcoded credentials, sensitive file path access (e.g., SSH or AWS keys), or unauthorized network operations were detected.- [Unverifiable Dependencies & RCE] (SAFE): No external packages are installed via pip or npm, and no remote scripts are downloaded or executed.- [Indirect Prompt Injection] (SAFE): While the skill ingests external data through the 'input_data' parameter in 'skill.py', it lacks the capabilities (such as file writing, network access, or shell execution) required for exploitation. (Ingestion: parameters['input_data'] in skill.py; Boundary markers: absent; Capability inventory: none; Sanitization: absent).- [Dynamic Execution] (SAFE): The modification of sys.path uses a local, relative path to load a framework directory, which is a standard operational pattern and does not involve untrusted input.- [Persistence & Privilege Escalation] (SAFE): No attempts to gain elevated privileges or maintain persistence on the host system were identified.
Audit Metadata