dialectic

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs spawning research subagents that perform web searches and fetch external pages (see Phase 1d "Full parallel research" and the Phase 3 example using --allowedTools web_search,web_fetch), so agents are expected to fetch and read open/public web content (potentially user-generated/untrusted) which can materially influence their arguments and downstream actions, creating risk of indirect prompt injection.