humanizer
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a transparent scoring engine for text analysis using local JavaScript logic. Analysis of the source code confirms no network operations, credential harvesting, or malicious command execution are performed.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-provided text. While this creates a surface for indirect prompt injection, the skill lacks the capabilities (such as network access or subprocess execution) to be exploited maliciously.
- Ingestion points: User text provided via stdin or local file reading in
src/cli.js. - Boundary markers: None explicitly defined in the input processing logic.
- Capability inventory: Local file system read access via
fs.readFileSyncand standard console output. No network or dynamic execution (eval/exec) capabilities were found. - Sanitization: The text is tokenized and processed for statistical metrics, providing a layer of separation from executable logic.
Audit Metadata