plan-design

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md:24-30 uses mcp__plugin_linear_linear__get_issue to fetch a Linear issue’s Description/Acceptance criteria (outsider-authored text), and SKILL.md:124-125 then injects that issue-derived “screen context from Linear issue” into Phase 2 image prompt strings sent to the LLM/image model at runtime.