content-voice
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted external data in the form of creator examples, published videos, and transcripts to build a voice profile.
- Ingestion points: Step 1 requires the agent to gather three to ten approved examples, such as finished posts, published videos, or transcripts, which may contain attacker-controlled content.
- Boundary markers: The instructions do not specify the use of delimiters or 'ignore' commands to separate the style evidence from the agent's instructions during the extraction phase.
- Capability inventory: The skill is capable of searching the private workspace and performing file system writes to create or update 'CONTENT_STYLE.md'.
- Sanitization: There is no evidence of sanitization or filtering of the source materials to prevent embedded instructions from influencing the profile extraction logic.
Audit Metadata