lp-optimizer

Pass

Audited by Gen Agent Trust Hub on Sep 6, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of URLs and HTML content for analysis.
  • Ingestion points: The Quick Start and Phase 1: Discovery sections in SKILL.md direct the agent to fetch and analyze external marketing pages.
  • Boundary markers: There are no explicit delimiters or safety instructions provided to prevent the agent from following instructions potentially embedded within the target HTML/web content.
  • Capability inventory: The agent is instructed to generate implementation code (HTML/CSS/JS) as part of Phase 4, which could be exploited if the source content contains malicious payloads.
  • Sanitization: No sanitization or filtering logic is specified for the ingested content.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 6, 2026, 07:12 AM
Security Audit — agent-trust-hub — lp-optimizer