legal-service-launch-checklist
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill relies on official Korean government and regulatory bodies for its data, including the National Law Information Center (law.go.kr), AI Support Desk (sw.or.kr), Personal Information Protection Commission (pipc.go.kr), KISA (kisa.or.kr), and the Fair Trade Commission (ftc.go.kr). These are well-known and trusted authoritative sources.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface as it processes user-provided service descriptions to generate its checklist.
- Ingestion points: User queries describing service type, billing models, and AI usage (SKILL.md).
- Boundary markers: None identified in the prompt templates.
- Capability inventory: Limited to text processing and document formatting via other vendor-specific tools (moai-writer, moai-officer).
- Sanitization: None explicitly mentioned.
- Assessment: Because the skill lacks high-privilege capabilities like network exfiltration, arbitrary file writing, or shell execution, the vulnerability to indirect injection is negligible.
- [COMMAND_EXECUTION]: No shell commands, dynamic scripts, or suspicious command patterns (like backtick injection) were detected in any of the skill files.
- [DATA_EXFILTRATION]: No network operations, hardcoded credentials, or sensitive path access were found. Mentions of external platforms like Supabase and AWS are restricted to legal residency context and do not involve data transmission.
Audit Metadata