soc2-gap

Pass

Audited by Gen Agent Trust Hub on Jul 20, 2026

Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [PROMPT_INJECTION]: The skill incorporates defensive instructions to disregard directives found within analyzed content. It also possesses a surface for indirect prompt injection. Ingestion points: Data provided via $ARGUMENTS and files read using Read, Grep, and Glob tools. Boundary markers: Explicit instructions in the Prompt Injection Safety Notice to treat embedded text as data. Capability inventory: Restrictive toolset with no network or code execution access. Sanitization: Instructions to redact credentials and sensitive tokens from reports.
  • [DATA_EXFILTRATION]: The skill audits sensitive infrastructure configurations but includes mandatory data protection rules that prevent the leakage of raw credentials or API keys in the final output.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 20, 2026, 02:11 PM
Security Audit — agent-trust-hub — soc2-gap