ad-creative

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). This skill ingests user-provided performance data (CSV/pasted/API output) via “Mode 2: Iterate from Performance Data,” then the LLM analyzes and uses that text to generate new ad copy, so an outsider can submit poison through the workflow’s required runtime input.

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 1.00). The skill includes explicit ad platform management commands (e.g., "google-ads campaigns create", "meta-ads campaigns list") and workflow steps to "Upload to platform" and manage campaigns. These integrations imply the ability to create/manage ad campaigns (including budgets/ad spend) via platform APIs — i.e., direct ad spend management, which qualifies as direct financial execution authority.