swap-voice
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes external video transcripts, which is a potential surface for indirect prompt injection. However, the risk is inherent to the skill's primary function and is mitigated by human-in-the-loop checkpoints during the voice selection and final review steps.
- Ingestion points: Video transcripts from existing Clueso projects or raw recordings (SKILL.md, Workflow step 2).
- Boundary markers: None mentioned for transcript content.
- Capability inventory: Voice shortlist generation, narration regeneration, and project export (SKILL.md, Workflow steps 3, 4, 7).
- Sanitization: Not explicitly defined in the instructions.
Audit Metadata