citation-claim-audit

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

• Third-party content exposure detected (high risk: 0.80). The skill's required workflow explicitly instructs the agent to "Locate cited evidence and, when needed, the raw PDF or experiment artifact" (SKILL.md step 3) and to use literature search artifacts (references to sota/search-strategy.md and storing sources/pdfs/), which indicates the agent will ingest external third-party documents whose content can directly affect verdicts and edits.