read-book
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted document content from local files and public-domain URLs (Project Gutenberg, Archive.org). This creates a surface for indirect prompt injection where instructions embedded in a book could influence the agent's behavior.\n
- Ingestion points: File system via Read tool and URLs via WebFetch (references/sources.md).\n
- Boundary markers: None identified.\n
- Capability inventory: Shell command execution for document tools and file writing (references/output-modes.md).\n
- Sanitization: None identified.\n- [PRIVILEGE_ESCALATION]: The skill instructions include high-privilege commands using sudo for updating and installing LaTeX components (references/output-modes.md).\n- [COMMAND_EXECUTION]: The skill invokes multiple command-line utilities for processing books, including pandoc, pdfinfo, pdftotext, ebook-convert, ocrmypdf, pdftk, mutool, and open.\n- [EXTERNAL_DOWNLOADS]: The skill fetches content from well-known sources like Project Gutenberg and Archive.org via WebFetch and suggests installing system dependencies using brew.
Audit Metadata