content-sanitization
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill defines a defensive framework for handling untrusted data, specifically addressing risks by establishing clear boundary markers and sanitization steps for external content ingested from GitHub or the web.\n- [SAFE]: It provides actionable guidance to prevent command injection and remote code execution by restricting the use of external inputs in sensitive execution sinks like eval(), exec(), and shell-based subprocesses.\n- [SAFE]: The guidelines include techniques to detect and remove obfuscation attempts, such as hidden Unicode characters or CSS styling designed to conceal text from human reviewers.
Audit Metadata