meta-diagram-triangulation

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

• Third-party content exposure detected (high risk: 0.85). The skill ingests {{ outputs.scan_repo | truncate(2000) }} into the render_plantuml and render_drawio sub-agent prompts; scan_repo is a runtime repo scan that can include outsider-authored free text (e.g., README/docs/issues/code comments) which is then embedded into the LLM context.