mql5-ea-expert
Pass
Audited by Gen Agent Trust Hub on Jul 10, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill contains comprehensive development standards for MQL5, including proper lifecycle management for indicator handles and thread-safe bar detection.
- [COMMAND_EXECUTION]: The skill includes code for a 'CSV Logger' using MQL5's
FileOpenandFileWritefunctions. These operations are restricted by the MetaTrader 5 platform to a sandboxed directory (MQL5/Files), preventing arbitrary file system access. - [PROMPT_INJECTION]: The skill includes 'Safety Rules' that instruct the AI to refuse dangerous trading configurations and enforce risk management (e.g., 'NEVER remove risk management'). These are internal guardrails designed to protect the user from financial loss rather than attempts to bypass platform security.
- [DATA_EXFILTRATION]: The specialization section mentions 'Telegram bot (via WebRequest)'. While
WebRequestcan transmit data, the MetaTrader 5 terminal requires users to manually whitelist specific URLs in the platform settings, which serves as an effective control against unauthorized data exfiltration. - [SAFE]: No obfuscation, hardcoded credentials, or unauthorized remote code execution patterns were identified across the provided files.
Audit Metadata