publication-grounded-solution-design

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime workflow requires searching and ingesting outsider-authored free text (e.g., “peer-reviewed papers… technical reports… credible industry case studies”) into the agent’s LLM context via the precedent/literature search and “build a precedent table” steps.