claudeception

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md "Step 2: Research Best Practices" explicitly requires the agent to perform web searches, "review top results and incorporate relevant information," and add a "References" section with source URLs, meaning the agent will fetch and interpret untrusted public web content (search results, docs, community posts) that can influence skill creation and subsequent actions.