skill-official-evaluation

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

• Third-party content exposure detected (high risk: 0.75). Required workflow runs scripts/official_evaluate.py which reads the target skill’s text files (e.g., SKILL.md, scripts/*.py, references/*.md) from a user-supplied --skill-dir, so if that target skill is outsider-authored its free-form prose is ingested into the evaluator and then rendered into the LLM-visible report.