master-yinguang

Pass

Audited by Gen Agent Trust Hub on May 10, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is a character-based knowledge agent for Buddhist studies. A comprehensive audit of all provided files, including instructions, sources, and metadata, revealed no malicious patterns, obfuscation, or unauthorized data access.
  • [COMMAND_EXECUTION]: The skill instructions mention the use of local utility scripts (scripts/cite.py, scripts/query.py) to facilitate citation verification and database queries. These tools are scoped to the agent's primary function of retrieval-augmented generation and do not exhibit signs of privilege escalation or remote code execution.
  • [PROMPT_INJECTION]: The skill employs strict behavioral constraints defined within <HARD-GATE> blocks, such as requiring CBETA citations for every doctrinal claim and prohibiting sectarian judgment. These instructions are designed to ensure factual grounding and ethical compliance rather than to bypass safety protocols.
  • [EXTERNAL_DOWNLOADS]: References to external domains like fojin.app are restricted to static documentation links and public scripture IDs. No dynamic fetching of executable code or unauthorized data exfiltration patterns were detected.
Audit Metadata
Risk Level
SAFE
Analyzed
May 10, 2026, 10:28 AM