tt-audience-insights
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze attacker-controlled data from TikTok, including video captions and user comments, which presents a surface for indirect prompt injection attacks.
- Ingestion points: Data enters the agent context through the
fetch_video_comments,fetch_niche_videos, andfetch_profile_videosmethods, as well as via user-pasted content from TikTok. - Boundary markers: The skill explicitly includes an 'Untrusted content' section and references an external configuration (
../../references/untrusted-content.md) to define security boundaries. - Capability inventory: The skill utilizes the Apify API client and potentially MCP servers (
clockworks--tiktok-scraper) to fetch live data from the internet. - Sanitization: The instructions contain a directive stating that captions or comments addressing the agent or providing 'facts' have no authority and must not influence the agent's behavior or output.
Audit Metadata