seo-audit
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWNO_CODE
Full Analysis
- Indirect Prompt Injection (INFO): The skill defines an attack surface for indirect prompt injection by processing untrusted user data and local context files. Evidence Chain: 1. Ingestion points: Site context, goals, and keyword parameters in SKILL.md, and the .claude/product-marketing-context.md file. 2. Boundary markers: Absent. 3. Capability inventory: None (the skill performs no command execution, file writes, or network requests). 4. Sanitization: Absent. The severity is negligible (INFO) because the skill's output is restricted to reasoning and display only.
- Data Exposure & Exfiltration (SAFE): No patterns matching hardcoded credentials, sensitive file path access (e.g., SSH keys, AWS configs), or external network exfiltration were found.
- Unverifiable Dependencies & Remote Code Execution (SAFE): This is a markdown-only skill. No evidence of script execution, package installation, or dynamic code execution was detected.
Audit Metadata