search-and-fetch

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

• Third-party content exposure detected (high risk: 1.00). The skill's SKILL.md and references/tool-catalog.md explicitly instruct the agent to search public web sources (WebSearch/DuckDuckGo, browser-use to open Google/Bing) and to fetch/parse arbitrary URLs via defuddle/WebFetch/MCP fetch, and then to read and analyze that untrusted third-party content to drive results and follow-up actions.