humanize
Pass
Audited by Gen Agent Trust Hub on Jul 12, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted input and reads from local files provided by the user for voice calibration, creating a surface for indirect prompt injection.
- Ingestion points: Commands, inline text, and writing samples provided via the
--sample <file>argument. - Boundary markers: The workflow lacks explicit delimiters to separate user-provided content from the agent's instructions.
- Capability inventory: Accesses file system (read-only) for samples and performs text generation; no network or write access is triggered by the processed data.
- Sanitization: No pre-processing sanitization of input text or file content is performed.
- [EXTERNAL_DOWNLOADS]: The skill instructions include an installation command
npx skills@latest add pavelsimo/skills. This command downloads the skill components using the platform's standard package manager and targets the author's specific skill repository.
Audit Metadata