unity-modding
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists of educational documentation and C# code snippets for game architecture. No malicious code, prompt injections, or obfuscation techniques were detected.\n- [INDIRECT_PROMPT_INJECTION]: The skill identifies architectural surfaces that ingest untrusted third-party mod data.\n
- Ingestion points: The LoadModConfig and LoadModCatalog functions process external JSON files and catalogs from local directories.\n
- Boundary markers: The documentation explicitly advises developers that they should 'never trust mod input'.\n
- Capability inventory: The proposed system has the capability to read local files and instantiate game objects based on external data.\n
- Sanitization: Recommends specific sanitization measures including MoonSharp instruction limits, restricted API surfaces (no System.IO), and registry-based validation for entity spawning.\n- [DYNAMIC_EXECUTION]: The guide discusses runtime execution of Lua scripts and IL code patching via Harmony. These are core functionalities for the stated purpose of game modding and are accompanied by necessary security mitigation strategies such as execution sandboxing and library whitelisting.
Audit Metadata