novel-creator
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a local Python script,
scripts/create_epub.py, during Phase 5 to generate the final ebook. This is an integral part of the skill's stated purpose of producing EPUB files from story content. - [EXTERNAL_DOWNLOADS]: The
scripts/create_epub.pyfile relies on theebooklibPython package. While this is an external dependency, it is a well-known library for ebook manipulation and is consistent with the skill's functionality. - [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it ingests untrusted user input through the
AskUserQuestiontool during Phase 1 (Story Setup) and Phase 2 (Interactive RPG). This input is used to generate story chapters and thetask_plan.mdfile, which are subsequently processed by the EPUB generator script. - Ingestion points: User concept input and choice selections in Phase 1 and 2.
- Boundary markers: None explicitly defined in the prompt instructions to delimit user-provided plot points from agent instructions.
- Capability inventory: The skill writes Markdown files and executes a local script (
create_epub.py) that reads these files. - Sanitization: The Python script uses standard regular expressions to convert Markdown structures to HTML but does not specifically sanitize the content for malicious injection into the EPUB reader environment. However, this risk is considered low and inherent to creative writing assistants.
Audit Metadata