rgpd

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONMETADATA_POISONING
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill fetches and analyzes untrusted content from external URLs (homepage, privacy policies, contact forms) using WebFetch. This data ingestion introduces a risk of indirect prompt injection, where an attacker could embed malicious instructions in a website to influence the agent's behavior or the audit's findings.\n
  • Ingestion points: SKILL.md specifies fetching homepage and privacy policy content.\n
  • Boundary markers: There are no instructions to delimit or ignore potential prompts within the fetched content.\n
  • Capability inventory: The skill is capable of writing Markdown and Word files to the local file system.\n
  • Sanitization: No sanitization of the external data is specified.\n- [METADATA_POISONING]: The skill documentation includes hardcoded absolute paths (C:/Users/cmano/claude-seo/...) for output files. This discloses a local username ('cmano') and assumes specific system configurations. Furthermore, there is a discrepancy between the metadata author ('Zythos Media') and the platform-provided author ('carlosuhart'), which may mislead users regarding the skill's origin.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 12:02 AM
Security Audit — agent-trust-hub — rgpd