find-sources

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

• Third-party content exposure detected (medium risk: 0.30). The runtime workflow in SKILL.md instructs downloading/probing candidate sources from external URLs (e.g., API endpoints, bulk files, and search pages) and then reading the returned HTML/CSV/text at step 2/3, so an outsider who can cause the workflow to probe attacker-controlled URLs or records could supply poison free text for the LLM to ingest.