puda-machine-vision-validation

Pass

Audited by Gen Agent Trust Hub on Jul 21, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious patterns or security risks were identified. The skill is designed as a safety guardrail for physical hardware interaction.\n- [PROMPT_INJECTION]: The instructions are strictly instructional and do not attempt to bypass safety constraints or override agent behavior.\n- [DATA_EXFILTRATION]: No exfiltration vectors or hardcoded credentials detected; explicitly warns against credential exposure.\n- [COMMAND_EXECUTION]: Describes physical validation workflows with a strict non-execution policy for validation-only requests.\n- [EXTERNAL_DOWNLOADS]: All referenced files and skills are part of the local environment or authorized PUDA ecosystem.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 21, 2026, 03:57 AM
Security Audit — agent-trust-hub — puda-machine-vision-validation