fix-pronunciations
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external transcript data and user-provided phonetic spellings, which represents a potential ingestion surface for untrusted content.
- Ingestion points: Video project transcripts and user-defined pronunciation strings.
- Boundary markers: No explicit delimiters are specified for separating project data from agent instructions.
- Capability inventory: Transcript searching and audio clip regeneration through the Clueso MCP interface.
- Sanitization: The skill does not describe specific sanitization or filtering logic for the ingested text data.
Audit Metadata