mt5-ea-coder
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill acts as a consumer for strategy specifications provided in JSON format. Ingestion points: Untrusted data enters via the strategy-spec-risk JSON input referenced in the Workflow and Decision Tree sections. Boundary markers: The skill uses formal JSON parsing and structured logic to separate specification data from generation instructions. Capability inventory: The skill generates MQL5 code which includes file I/O (logging) and network communication (WebRequest) capabilities. Sanitization: The skill includes 'Pre-Generation Validation' and 'Input Validation' checklists to ensure data consistency and validity before code generation.\n- [DATA_EXFILTRATION]: Provides standard implementation patterns for Discord and Telegram notifications. These use user-configured webhooks and bot tokens to transmit trade alerts, which is an expected feature for trading software and uses well-known services.\n- [DYNAMIC_EXECUTION]: Includes documentation and templates for utilizing the MT5 ONNX integration, allowing Expert Advisors to run local machine learning models for inference within the MetaTrader environment.
Audit Metadata