voice-extractor
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill accepts unverified user writing samples to analyze communication DNA, which introduces an indirect prompt injection surface.
- Ingestion points: Writing samples from emails, Slack messages, transcripts, and articles are collected into the agent context as described in
SKILL.mdandSKILL-OC.md. - Boundary markers: There are no structural boundary markers or security instructions specified to enclose or isolate the text samples from the rest of the prompt logic.
- Capability inventory: No active tools, network calls, file modifications, or code executions are defined in
SKILL.md,SKILL-OC.md, orVOICE-GUIDE-TEMPLATE.md. - Sanitization: The input is checked for character count and context adequacy but does not undergo sanitization or parsing to filter potential injection payloads.
Audit Metadata