tech-decision

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly invokes the "dev-scan" step to collect community opinions from public, user-generated sites (Reddit, Hacker News, Dev.to, Lobsters), which the agents will read and synthesize, exposing the agent to untrusted third-party content that could carry indirect prompt injection.