accelint-skill-prose

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime workflow is driven by text the user provides and by the local skill-folder artifact files it reads (e.g., root SKILL.md and references/*.md), so outsider-authored free text can be injected into those editor inputs to influence the LLM’s prompt content.