ai-agent-governance
Pass
Audited by Gen Agent Trust Hub on May 10, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill contains phrases typically associated with prompt injection, such as 'ignore previous instructions' and 'disregard instructions'. These are contained within a defensive regex list (
injection_patterns) intended to detect and block attacks as part of its 'Risk Management' pillar. They do not constitute a malicious attempt to override agent behavior. - [DATA_EXFILTRATION]: The skill defines patterns for identifying sensitive data (SSN, credit cards, API keys). These patterns are used for detection and redaction (PII protection) via Microsoft Presidio, representing a security best practice rather than an exfiltration attempt.
- [EXTERNAL_DOWNLOADS]: The skill references standard, well-known libraries for security and observability, including
presidio-analyzer,presidio-anonymizer,opentelemetry, andprometheus_client. These are reputable open-source tools. - [COMMAND_EXECUTION]: The code samples include logic for handling 'execute' permissions within an RBAC system and tool guardrails. These are architectural examples for controlling access and do not execute arbitrary shell commands.
Audit Metadata