repurpose-into-a-post
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, including articles, transcripts, and long documents, to generate social media content. An attacker could embed malicious instructions within these source materials to influence the agent's output or tool usage.
- Ingestion points: Sources identified in
SKILL.mdinclude articles, newsletters, transcripts, videos, and long documents provided by the user. - Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded commands within the processed data.
- Capability inventory: The skill utilizes high-impact tools including
publish_postandschedule_postwhich have direct side effects on social media platforms. - Sanitization: The skill includes a robust mitigation requiring the agent to show a preview and obtain a one-time code from the user before finalizing a post.
- [DATA_EXFILTRATION]: The skill documentation notes that when the VoiceMoat connector is enabled, post text and account data are sent to the vendor domain
voicemoat.com. This is a standard vendor-provided service function for this skill, but it involves transferring potentially sensitive draft content and metadata to an external infrastructure.
Audit Metadata