brief

Pass

Audited by Gen Agent Trust Hub on Apr 23, 2026

Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill explicitly prompts users for and stores personally identifiable information (PII), specifically client email addresses, within the persistent local filesystem in the 'location.brief.md' files.
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface:
  • Ingestion points: The skill reads and processes data from various external sources, including 'LocalSEOData' tool outputs (business profiles, reviews, citation audits) and results from scheduled tasks.
  • Boundary markers: There are no instructions defining delimiters or boundary markers to distinguish between trusted instructions and untrusted data retrieved from external SEO audits or GBP (Google Business Profile) content.
  • Capability inventory: The skill is capable of performing filesystem write operations (creating/updating brief directories) and executing multiple automated SEO analysis tools.
  • Sanitization: The instructions lack requirements for sanitizing or validating external content before it is incorporated into the persistent brief files or used to determine 'Next Actions'.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 23, 2026, 12:16 AM
Security Audit — agent-trust-hub — brief