cutagent-speech-cleanup
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted transcript data, creating a potential surface for indirect prompt injection.
- Ingestion points: Transcript segments processed in bounded windows in SKILL.md.
- Boundary markers: Absent; no delimiters are used for transcript content.
- Capability inventory: References to CutAgent CLI and SDK tools in SKILL.md suggest capabilities for file editing.
- Sanitization: Absent; no sanitization of input text is mentioned.
- [NO_CODE]: The skill consists only of natural language instructions and lacks executable code.
Audit Metadata