epub-translator

Pass

Audited by Gen Agent Trust Hub on Jul 9, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements security best practices for handling file archives. The provided script, scripts/translate_epub.py, includes logic to prevent path traversal attacks by validating that all extracted files remain within the designated work directory and rejecting entry paths containing parent directory references ('..').
  • [PROMPT_INJECTION]: The skill handles untrusted data from user-provided EPUB files, which represents an indirect prompt injection surface. This is addressed through structured data boundaries and specific instructions:
  • Ingestion points: Text content is extracted from XHTML and XML documents within the EPUB into chunks/*.jsonl files (scripts/translate_epub.py).
  • Boundary markers: The agent is provided with a strict JSONL schema and explicit instructions to only fill the 'translation' field, preserving all other keys and tags.
  • Capability inventory: The utility is limited to local file system operations, including reading, writing, and ZIP management; no network access or external command execution is performed.
  • Sanitization: The script performs automated checks on the resulting JSONL files to ensure ID consistency and the integrity of EPUB placeholder tags.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 9, 2026, 07:27 AM
Security Audit — agent-trust-hub — epub-translator