@1247/trade-simulator

Pass

Audited by Gen Agent Trust Hub on Mar 15, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill ingests untrusted market data and user-provided scenarios that are interpolated into LLM system and user prompts across several scripts. This architectural pattern creates a surface for indirect prompt injection attacks where malicious data could influence the simulation logic or outcome.\n
  • Ingestion points: Market data dictionary and scenario strings in mirofish_engine.py, simulate.py, and simulation_runner.py.\n
  • Boundary markers: Absent; data is merged directly into f-string templates without protective delimiters (e.g., XML tags) or instructions to ignore embedded commands.\n
  • Capability inventory: The skill possesses extensive capabilities for LLM-driven reasoning, automated reporting, and interactive interview responses.\n
  • Sanitization: No validation or escaping of external data is performed before it is interpolated into the prompts.\n- [EXTERNAL_DOWNLOADS]: The skill communicates with the OpenRouter API for LLM reasoning and generation services. These network operations use standard Python libraries and target a well-known service provider essential for the skill's primary function.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 15, 2026, 04:50 PM