meta-skill-builder
Warn
Audited by Snyk on Aug 19, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (medium risk: 0.30). SKILL.md의 Phase 1.5 Research는 사용자 요청(Phase 1 Requirements) 후 WebSearch 결과 및 Context7 MCP로 수집한 외부 자료를 인메모리에 저장하고 Phase 3의
## 출처에 인용하는 런타임 경로가 포함되어, 제3자가 게시한 무료 텍스트(검색 결과/라이브러리 문서)가 LLM에 간접 주입될 수 있습니다.
MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).
- Potentially malicious external URL detected (high risk: 0.70). The skill mandates runtime WebSearch/Context7 fetches and "출처 보존" of external sites (e.g., nts.go.kr) whose fetched content and quotes are preserved in-memory and injected into Phase 3/skill prompts, so these external domains are used at runtime to influence the agent's instructions.
Issues (2)
W011
MEDIUMThird-party content exposure detected (indirect prompt injection risk).
W012
MEDIUMUnverifiable external dependency detected (runtime URL that controls agent).
Audit Metadata