beopsuny

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches legal information and datasets from external sources to provide accurate answers:
  • It clones and pulls legal data (statutes, administrative rules, and precedents) from the legalize-kr organization on GitHub (e.g., github.com/legalize-kr/legalize-kr).
  • It utilizes the Beopmang API (api.beopmang.org) for real-time discovery of administrative rules and legal interpretations.
  • It retrieves knowledge manifests and assets from the author's public GitHub repository (github.com/sungjunlee/beopsuny-knowledge).
  • [INDIRECT_PROMPT_INJECTION]: Because the skill is designed to analyze untrusted user data, such as contracts or legal memos, it includes robust defensive instructions:
  • The references/self-verification.md file contains a "Retrieved Content Trust" section that explicitly directs the agent to treat all ingested documents as data for analysis, not as instructions.
  • It mandates that the agent ignore phrases like "ignore previous instructions" or "delete disclaimers" found within analyzed documents, preserving the skill's internal safety and verification protocols.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 06:28 AM
Security Audit — agent-trust-hub — beopsuny