valyu

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 explicitly exposes web-sourced, arbitrary URLs via /v1/search (web searches), /v1/contents (POST /v1/contents with urls[]), and /v1/answer (which blends search results into AI-generated answers) — meaning untrusted public web content is fetched, extracted, and used to drive agent outputs and decisions.