skills/steveclarke/dotfiles/humanizer/Gen Agent Trust Hub

humanizer

Pass

Audited by Gen Agent Trust Hub on Mar 8, 2026

Risk Level: SAFEPROMPT_INJECTIONNO_CODEEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through user-provided text.
  • Ingestion points: The skill is designed to read and process untrusted external text at runtime for the purpose of editing and humanizing content.
  • Boundary markers: There are no specific delimiters or instructions provided to the agent to separate the source data from the skill's own operational instructions.
  • Capability inventory: The agent is granted extensive file-system permissions, including Read, Write, Edit, Grep, and Glob tools.
  • Sanitization: No input sanitization or validation logic is present to identify or neutralize malicious instructions embedded within the text provided by users.
  • [NO_CODE]: The skill consists exclusively of instructional markdown and metadata, with no executable scripts, binaries, or software dependencies included.
  • [EXTERNAL_DOWNLOADS]: The skill includes a reference link to Wikipedia's documentation on AI writing patterns as a knowledge source for its instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 8, 2026, 05:21 PM