humanizer

Pass

Audited by Gen Agent Trust Hub on Sep 11, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted text from users and external files to apply stylistic changes, which creates a surface for indirect prompt injection that could influence the agent's behavior during the editing process.\n
  • Ingestion points: The skill explicitly ingests content via pasted text in conversation and through reading external files on the system in 'File mode'.\n
  • Boundary markers: There are no specific delimiters or 'ignore' instructions provided to the agent to prevent the processed text from being interpreted as instructions rather than data.\n
  • Capability inventory: The skill is granted Read, Write, Edit, Grep, and Glob permissions, which allows it to modify the filesystem based on the interpreted content.\n
  • Sanitization: The skill focuses on stylistic sanitization (e.g., removing banned words) but does not include mechanisms to filter or escape instructions that might trigger unintended agent actions.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 11, 2026, 11:19 PM
Security Audit — agent-trust-hub — humanizer