vsl-script

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFE
Full Analysis
  • [DATA_EXPOSURE_EXFILTRATION]: The skill manages brand profiles and configuration in ~/.claude/ad-profiles/ and saves generated content to a local directory (defaulting to ~/Documents/ad-copy/). This behavior is used for cross-skill data persistence and does not involve external transmission.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from user-provided testimonials and brand stories stored in local markdown files.\n
  • Ingestion points: Brand profile markdown files (~/.claude/ad-profiles/*.md) and interview responses.\n
  • Boundary markers: Absent explicitly for file-based ingestion.\n
  • Capability inventory: Performs local file-system read and write operations.\n
  • Sanitization: The skill includes a 'do not say' list to filter specific content, though it lacks specialized sanitization for preventing instructions embedded in the input data.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 08:24 AM
Security Audit — agent-trust-hub — vsl-script