novel-style-extract
Pass
Audited by Gen Agent Trust Hub on May 14, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill operates by processing literary text provided by the user and generating structured templates. No evidence of malicious intent or dangerous capabilities was found.
- [DATA_EXPOSURE]: The skill requests access to local files (.txt, .md, .epub) for analysis. However, it explicitly states in the 'Boundaries' section that it does not access unauthorized files and relies on paths specifically designated by the user. There are no network operations present to facilitate data exfiltration.
- [INDIRECT_PROMPT_INJECTION]: As the skill ingests untrusted external data (novel text), there is a theoretical surface for indirect prompt injection. This is effectively mitigated by the 'User Confirmation and Adjustment' step (Step 5 of the workflow), which mandates that the user reviews and confirms extracted 'Ironclad Rules' and content before any files are written to the directory. This human-in-the-loop verification reduces the risk of malicious instructions from the source text being adopted into the agent's knowledge base.
- [COMMAND_EXECUTION]: No shell commands, dynamic context injections (!command), or subprocess calls were identified in the instructions or references.
Audit Metadata