book2skill
Pass
Audited by Gen Agent Trust Hub on Jun 18, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it is designed to ingest and process arbitrary external text from books to generate new agent instructions. A maliciously crafted book could theoretically contain hidden commands intended to influence the distillation process or the resulting skill outputs.
- Ingestion points: The skill reads user-provided PDF, EPUB, and TXT files.
- Boundary markers: The instructions do not define specific delimiters or 'ignore embedded instructions' warnings for the input book text.
- Capability inventory: The skill generates new SKILL.md files and spawns sub-agents via the Agent tool to process content.
- Sanitization: While the methodology includes conceptual verification (Triple Verification), it does not implement technical sanitization for prompt injection patterns in the source text.
- [COMMAND_EXECUTION]: The skill performs expected file system operations (reading source files and writing to the '/books/' directory) and utilizes agent tools to orchestrate the parallel execution of extraction sub-agents.
Audit Metadata