docling

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

• Third-party content exposure detected (high risk: 0.90). Yes — the skill instructs the agent to fetch and parse arbitrary web pages (e.g., the SKILL.md examples using docling "" --from html and the workflow "For web content extraction: Use docling """) and then read the resulting output, which means it ingests untrusted public content from arbitrary URLs.