codex

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly enables built-in web search (see "Web Search Note" in SKILL.md and the web-search examples in references/advanced-patterns.md and references/examples.md such as "--enable web_search_request" and "Find latest JWT security best practices") and instructs the agent to read and act on public web content to inform research, design, and code-editing actions, so untrusted third-party content can influence subsequent tool use and decisions.