planning-with-files

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly instructs the agent to perform "search, browse, or read operations" and to write/re-read "browser returned data" and keep URLs (see Core Pattern / Critical Rules and References/Principles), which indicates the agent will ingest arbitrary public web content that can influence decisions and tool use.