research

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

• Third-party content exposure detected (high risk: 0.90). The skill's Investigation workflow (see SKILL.md "Investigation" cognitive phases and "Gather" step) explicitly instructs spawned subagents to perform year‑pinned WebSearch → WebFetch on top results and quote verbatim from public sites, and those synthesized third‑party sources are then parsed and used to drive Storage and follow-up actions, which exposes the agent to untrusted web content that could carry indirect prompt injection.