skill-sentinel

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md’s required runtime workflow runs scripts/scanner.py to discover skills under SKILL_SEARCH_PATHS and then analyzers read outsider-authored SKILL.md, references/*.md, and scripts/*.py from those directories into LLM-readable strings (e.g., read_text(...)), which can include free-form text from other parties.