unity-vrc-udon-sharp

Pass

Audited by Gen Agent Trust Hub on May 5, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill serves as a technical reference and template library for VRChat's UdonSharp scripting system. The content is educational and functional.
  • [EXTERNAL_DOWNLOADS]: All referenced URLs point to official VRChat domains (creators.vrchat.com, feedback.vrchat.com, ask.vrchat.com) or the official community GitHub organization (github.com/vrchat-community). These are trusted resources for the intended use case.
  • [COMMAND_EXECUTION]: The skill includes shell and PowerShell hooks (validate-udonsharp.sh and validate-udonsharp.ps1) designed to run as post-tool-use validation steps. Analysis confirms these scripts strictly perform regex-based static analysis to detect SDK constraint violations (e.g., usage of blocked generics or async/await) and do not contain malicious payloads.
  • [DATA_EXFILTRATION]: No patterns indicative of data theft were found. Technical delimiters like U+2029 and Base64 encoding are used legitimately within the context of VRChat SDK limitations for networking and texture loading.
  • [PROMPT_INJECTION]: Instructions labeled as 'IMPORTANT' or 'CRITICAL' in the rules files are operational constraints for the AI agent to prevent it from generating non-compiling UdonSharp code, rather than attempts to bypass safety filters.
Audit Metadata
Risk Level
SAFE
Analyzed
May 5, 2026, 08:19 AM