claude-context-management

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly references and demonstrates ingesting web search results (e.g., tools=[{"type":"web_search_20250305","name":"web_search"}]) and discusses "web search workflows (50+ searches)" and preserving/clearing tool results, meaning the agent will read and act on public, untrusted web content.