lfpdppp

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFEDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill audits external websites by fetching content from user-provided URLs. It lacks boundary markers or explicit instructions to sanitize or ignore instructions embedded within the fetched content.
  • Ingestion points: Website content fetched from the provided <url> argument.
  • Boundary markers: Absent; the instructions do not specify delimiters for external content.
  • Capability inventory: File system write access (Markdown and DOCX generation), network access to fetch audit data.
  • Sanitization: Absent; no escaping or filtering of external content before processing.
  • [DATA_EXPOSURE]: The skill instructions hardcode a specific local directory structure for saving reports, which leaks a local username.
  • Evidence: The skill specifies saving reports to C:/Users/cmano/claude-seo/[cliente]/, exposing the username cmano and the host's directory layout.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 12:02 AM
Security Audit — agent-trust-hub — lfpdppp