blog-translate
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill implements security best practices for file handling, including resolving paths against the project root and explicitly rejecting directory traversal and symlinked paths.
- [SAFE]: Input validation is performed on language codes and file types (rejecting binary files) to prevent processing unexpected data.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted blog content (markdown, MDX, or HTML) as its primary input, creating a potential surface for indirect instructions.
- Ingestion points: Phase 1 reads user-provided files after resolving paths against the project root.
- Boundary markers: Content is processed and passed to the translator agent without explicit structural delimiters or instructions to ignore embedded commands.
- Capability inventory: The skill has the ability to spawn sub-agents (Phase 4) and write localized content to the local filesystem (Phase 5).
- Sanitization: Robust path sanitization and file-type validation are implemented to prevent system-level traversal or execution of unauthorized files.
Audit Metadata