unocss

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

• Third-party content exposure detected (high risk: 0.70). The documentation (references/core-extracting.md) explicitly shows content.inline can call async functions like (await fetch('https://example.com')).text(), meaning the UnoCSS pipeline can fetch arbitrary public URLs and then read/interpret that remote text as part of its extraction workflow, exposing the agent to untrusted third‑party content.