content-strategy

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

• Third-party content exposure detected (high risk: 0.80). The skill's required "Customer Research for Content" workflow (Step 2) explicitly tells the agent to "mine" community forums (e.g., Reddit, Slack, Discord) and competitor product reviews — untrusted, user-generated third-party sources — and to use those findings to choose topics and drive content decisions, which clearly exposes the agent to content that could carry indirect prompt-injection instructions.