godot-skill-judge

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [Indirect Prompt Injection] (HIGH): The skill's primary purpose is to process untrusted data from external skill folders.
  • Ingestion points: Scans all files and folders in a skill library, specifically reading SKILL.md files and .gd script files.
  • Boundary markers: Absent. There is no evidence of delimiters or instructions to ignore embedded commands within the files being validated.
  • Capability inventory: Utilizes the Godot CLI for syntax checking, which involves system-level process execution.
  • Sanitization: Absent. There is no mention of filename or content sanitization before processing.
  • [Command Execution] (MEDIUM): The skill triggers the Godot CLI (godot) to validate scripts. If the file paths or script names of the skills being checked are not properly escaped, an attacker could use malicious filenames (e.g., ; rm -rf / ;.gd) to achieve command injection on the host system running the judge.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 03:33 AM