antithesis-research

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

• Third-party content exposure detected (high risk: 0.80). The skill's "Prerequisites and Scoping" (SKILL.md) explicitly asks the user for external references (paths or URLs) and the workflow/instructions in SUT discovery, property discovery, and evaluation (e.g., references/sut-discovery.md and references/property-discovery.md) require agents to consult and read those user-named external references, meaning the agent will fetch and interpret arbitrary third‑party content that can influence decisions.