ultra-humanizer
Pass
Audited by Gen Agent Trust Hub on Oct 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-supplied text for rewriting, creating a vulnerability surface where instructions embedded in that text could influence an agent with file system access.
- Ingestion points: The skill ingests raw text from external sources for the 'Diagnose' and 'Rewrite' steps of its core process in SKILL.md.
- Boundary markers: Absent. The instructions rely on stylistic constraints (preserve facts, change only style) rather than technical delimiters or explicit warnings to ignore embedded instructions.
- Capability inventory: The skill is enabled with broad file system access tools, specifically Read, Write, Edit, Grep, and Glob.
- Sanitization: Absent. There is no defined process for sanitizing, escaping, or validating the input text before the agent processes it in its local context.
Audit Metadata