humanizer

Warn

Audited by Gen Agent Trust Hub on Feb 19, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • EXTERNAL_DOWNLOADS (MEDIUM): The README recommends installation via git clone from a non-whitelisted GitHub repository (https://github.com/blader/humanizer.git). This falls outside the Trusted External Sources scope and constitutes an unverifiable dependency installation.
  • PROMPT_INJECTION (LOW): The skill possesses an Indirect Prompt Injection surface as it is designed to process untrusted user input to 'humanize' text.
  • Ingestion points: Users are instructed to paste text or use the template Please humanize this text: [your text] (README.md).
  • Boundary markers: None specified in the documentation to prevent the model from obeying instructions embedded within the text to be humanized.
  • Capability inventory: Primarily text transformation; however, the lack of sanitization on input that is fed directly into a transformation prompt is a known vector for indirect attacks.
  • Sanitization: No evidence of input filtering or escaping is provided in the skill documentation.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 19, 2026, 11:27 PM