omega-prompt-forge

Fail

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: HIGHPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [Prompt Injection] (HIGH): The file references/omega-promptforge.md contains high-severity jailbreak-style patterns. It uses pseudo-code structures (e.g., class UltraPromptEngine, activate_all_layers) and explicit instructions to 'break cognitive boundaries', 'allow creative rule bending', and 'recursive optimization until limit' to override base model safety protocols.
  • [Command Execution] (HIGH): The skill workflow in SKILL.md explicitly attempts to write files to a hardcoded, absolute local path: /Users/wisewong/Documents/Developer/prompts/<task-slug>/prompt.md. This bypasses standard sandboxing expectations and targets a specific user's directory, which could be exploited to modify sensitive system files or exfiltrate data if the path is redirected to configuration or credential directories.
  • [Obfuscation] (MEDIUM): The 'OmegaPromptForge' component uses technical metaphors and pseudo-executable code blocks to hide its intent to override agent behavior behind a facade of 'advanced cognitive architecture', which is a common tactic to evade automated safety filters.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 17, 2026, 06:40 PM