explore

Fail

Audited by Gen Agent Trust Hub on Apr 23, 2026

Risk Level: CRITICAL
Full Analysis
  • [SAFE]: The skill uses various search tools (LSP, ast-grep, exa MCP) to gather information. It delegates complex searching to independent agents to keep the primary context clean.
  • [SAFE]: The skill includes a blocklist in references/content-farm-domains.md to prevent the agent from trusting or citing low-quality scrapers and content farms. The URLs flagged as malicious by automated scanners (such as askcodez.com, isolution.pro, and higithub.com) are correctly placed in this blocklist for exclusion, which is a security best practice for web-enabled agents.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from web searches and codebase files. This is mitigated by the instruction to use AI semantic judgment to filter out filler text and keyword-stuffed content, as well as the explicit use of a domain blocklist to avoid known low-quality sources.
Recommendations
  • Contains 3 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Apr 23, 2026, 04:04 AM
Security Audit — agent-trust-hub — explore