senior-computer-vision
Pass
Audited by Gen Agent Trust Hub on Apr 5, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns or security vulnerabilities were identified in the analyzed files. The skill adheres to established development workflows and includes protective instructions against data-driven prompt injection.
- [COMMAND_EXECUTION]: The skill involves the execution of local scripts and standard command-line interfaces (CLIs) for machine learning frameworks to perform model training and optimization tasks. This behavior is aligned with the skill's engineering focus.
- [PROMPT_INJECTION]: The instructions specifically include a safety rule to prevent the agent from interpreting untrusted dataset annotations as executable instructions, mitigating risks of indirect prompt injection.
Audit Metadata