humanizer
Pass
Audited by Gen Agent Trust Hub on Aug 16, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection from user-provided data. 1. Ingestion points: User-provided text for humanization and writing samples for voice calibration (SKILL.md, Steps 1 and 5). 2. Boundary markers: Absent; the skill does not use delimiters or instructions to ignore embedded commands in the input. 3. Capability inventory: The agent has access to Bash, Write, Edit, Read, Grep, Glob, and AskUserQuestion. 4. Sanitization: None; input is processed and passed to calibration sections as-is.
- [OBFUSCATION]: Although zero-width characters were detected in scripts/humanizer_metrics/markers.py, they are part of a regex used for detecting AI-generated text patterns. This is a legitimate functional use case and not a security threat.
Audit Metadata