anthropic-docs

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

• Third-party content exposure detected (high risk: 0.90). This skill's documentation explicitly documents and demonstrates use of web-search and web-fetch tools (e.g., references/about-claude/models/whats-new-claude-4-6.md and the Agents & Tools/tool-use pages referenced in SKILL.md) which fetch open/public web content that the agent is expected to read and use to drive tool calls and actions, exposing it to untrusted third‑party content that could contain indirect prompt injection.