wjs-translating-subtitles
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-supplied content (SRT files and transcripts) to perform translations and formatting. While this is a common attack surface for LLM-based tools, the risk is inherent to its primary function.
- Ingestion points: User-provided source-language SRT files and transcript text used for translation.
- Boundary markers: The skill instructions do not specify explicit delimiters or "ignore embedded instructions" warnings for the subtitle content.
- Capability inventory: The skill is restricted to text processing and generating subtitle files; it does not contain network operations or arbitrary command execution.
- Sanitization: The skill focuses on translation quality and punctuation formatting without explicit sanitization of the input text content.
Audit Metadata