work-with-creators
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill represents an indirect prompt injection surface as it is designed to ingest and analyze untrusted content from various social media platforms.
- Ingestion points: External data enters the agent context via the
gooseworks:data_call_providertool (fetching TikTok, Instagram, YouTube, LinkedIn, and X content) and through the host browser's search and profile page views. - Boundary markers: The skill relies on structured report formats (e.g.,
creators/<D>/report.md) and evidence labeling ("public-page read") to distinguish between retrieved data and instructional context. - Capability inventory: The skill is capable of writing files to the workspace and generating messaging drafts, but it does not possess arbitrary command execution or persistence capabilities. It uses a
spend-gateto manage API costs and aclaims-gateto validate marketing claims. - Sanitization: The instructions mandate that marketing claims must be cleared through a dedicated
claims-gateskill before being included in briefs, providing a layer of validation for external data.
Audit Metadata