arize-compliance-audit
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted codebase data during its audit phase, which creates a potential surface for indirect prompt injection. * Ingestion points: Systematically scans the user's entire codebase for compliance evidence and gaps (SKILL.md, Phase 1). * Boundary markers: No specific boundary markers or instructions to ignore embedded commands are defined for the scanned code content. * Capability inventory: The skill possesses read-file capabilities, write-file capabilities for reports in /tmp, and code modification capabilities during remediation (SKILL.md, Phase 2 and 3). * Sanitization: The skill performs pattern matching rather than code execution on the scanned content, and all remediation actions require confirmation via the AskUserQuestion tool.
- [DATA_EXFILTRATION]: The skill performs targeted searches for sensitive data, including PII (emails, SSNs) and hardcoded credentials, to identify compliance and security gaps. * Evidence: This scanning is a core diagnostic feature necessary for GDPR and HIPAA compliance verification and is performed locally for auditing purposes (SKILL.md, Phase 1, Domains B and C).
- [EXTERNAL_DOWNLOADS]: The skill references multiple official regulatory documentation sites and recommends the installation of well-known security libraries. * Evidence: Provides links to official domains for NIST, ISO, and the European Union (SKILL.md, Reference links). * Evidence: Suggests the installation of established packages such as guardrails-ai and presidio-analyzer for remediation (SKILL.md, Phase 3).
Audit Metadata