chinese-novelist
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a bundled Python helper script,
scripts/check_chapter_wordcount.py, to verify that generated chapters meet the 3,000-5,000 word requirement. The script relies on standard Python libraries (os,re,sys) to perform local regex-based character counting and does not initiate network connections. - [INDIRECT_PROMPT_INJECTION]: The skill operates as a multi-step pipeline (Phases 0-4) that ingests user-provided story ideas and creative preferences. It manages this attack surface using strict markdown templates, structural outlines, and 'AI flavor removal' instructions which help constrain the generated output within the intended creative boundaries.
- [DATA_EXPOSURE]: The skill implements a persistence mechanism via
user-preferences.jsonto remember genre favorites and writing styles across sessions. It also creates project-specific JSON files to track progress. These operations are limited to the skill's working directory and do not access sensitive system paths, environment variables, or credentials.
Audit Metadata