humanizer
Pass
Audited by Gen Agent Trust Hub on May 11, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill's primary purpose is text processing and editing. While it processes untrusted text from users (ingestion point), which creates a surface for indirect prompt injection, the instructions are strictly limited to stylistic rewriting and pattern matching. No boundary markers or sanitization are present, but the restricted toolset (Read, Write, Edit, Grep, Glob) and the specific nature of the task mitigate risks.
- [EXTERNAL_DOWNLOADS]: The skill references Wikipedia's educational resources and a GitHub repository for installation. These are well-known and trusted services. There are no automated remote code execution patterns or downloads of executable content by the agent.
- [SAFE]: No obfuscation, hidden URLs, hardcoded credentials, or persistence mechanisms were found. The skill instructions stay within the scope of a text editor without attempting to escalate privileges or perform unauthorized network operations.
Audit Metadata