rn-cover-skill

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

• Third-party content exposure detected (medium risk: 0.30). In RN Cover Skill’s runtime workflow, outsider-provided inputs (label and title) are inserted directly into the LLM’s composed prompt/config for built-in ImageGen (via the {theme}/prompt assembly) and then used for SVG text via scripts/compose_cover.py, so the agent ingests free text before any “specific item” selection step.