localize-el
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external source text for translation, which constitutes an attack surface for indirect prompt injection.
- Ingestion points: Untrusted source text provided for translation as described in the Pre-Translation Setup section of SKILL.md.
- Boundary markers: The skill explicitly instructs the agent to preserve {variables} and ICU plural syntax, acting as a structural boundary.
- Capability inventory: The skill possesses no capabilities for file writing, network access, or shell execution across its instructions.
- Sanitization: No active sanitization is performed, as the skill focus is on legitimate text transformation and preservation of technical placeholders.
- [SAFE]: No malicious patterns, obfuscation, or unauthorized data access were identified. The instructions are focused on legitimate linguistic and localization tasks.
Audit Metadata