novel-editing
Pass
Audited by Gen Agent Trust Hub on Mar 3, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill contains only instructional text for guiding an AI's novel-editing behavior. No executable scripts, system commands, or network-enabled operations were detected.
- [DATA_EXPOSURE]: File access is restricted to relative paths within a project directory (e.g.,
draft/chapters/,bible/,reports/). These are standard for creative writing workflows and do not target sensitive system locations or user credentials. - [PROMPT_INJECTION]: The skill identifies a data ingestion surface by reading user-provided novel drafts (
draft/chapters/ch-*.md). While this is a theoretical surface for indirect prompt injection, the skill lacks the dangerous capabilities (such as shell execution or external network requests) required to exploit such an injection maliciously.
Audit Metadata