skill-installer

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

• Third-party content exposure detected (high risk: 0.85). Outsider-authored free text from a third-party skill is fetched at runtime (Step 2) and then ingested into the agent context as readable prose when the installer displays the full raw SKILL.md (Step 3) and runs skills-qa/trust checks (Steps 4–5), enabling indirect prompt-injection via that outsider SKILL.md content.