clean-user-facing-text

Pass

Audited by Gen Agent Trust Hub on Nov 13, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides a deterministic text cleaning utility focused on Unicode hygiene. Analysis of the scripts (clean_text.py, inspect_text.py, and text_unicode.py) shows they only process text passed as arguments or via stdin, performing character-level filtering and normalization.
  • [SAFE]: While the skill contains logic for handling Unicode homoglyphs and zero-width characters, this is the intended primary purpose of the skill (cleaning text) rather than an attempt to obfuscate malicious payloads. The text_unicode.py script identifies these characters for removal or replacement to ensure text hygiene.
  • [SAFE]: The skill instructs the agent to preserve facts, code, and identifiers, minimizing the risk of accidental data corruption or indirect injection when processing mixed content files.
  • [SAFE]: All file operations are localized to the input/output paths provided by the user, and the skill uses safe_write_bytes in common.py to prevent symlink attacks and ensure atomic writes with backups.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 13, 2026, 05:37 PM