ai-agent-procurement-objections-on-commercials

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

• Third-party content exposure detected (high risk: 0.75). The skill’s workflow consumes the “recorded objection, buyer rationale, and approved trade boundaries” from the buyer/proposal inputs (outsider-authored buyer words), which are free-form text that can be inserted into the agent’s LLM context via the Domain Inputs/Workflow steps.