cub-strategy

Pass

Audited by Gen Agent Trust Hub on Jun 28, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: Indirect prompt injection surface identified in the LLM decision-making process.
  • Ingestion points: The strategy script cub-producer.py fetches market data, including instrument names and asset contexts, from the Hyperliquid exchange using MCP tools such as market_get_asset_data and market_list_instruments.
  • Boundary markers: The LLM decision prompts in runtime-long.yaml, runtime-short.yaml, and runtime-preipo.yaml interpolate this external signal data using {{signal_...}} placeholders without explicit delimiters or instructions to treat the content as untrusted data.
  • Capability inventory: The skill has the OPEN_POSITION capability, enabling it to execute financial trades based on LLM evaluations.
  • Sanitization: String data retrieved from the exchange (such as asset names or metadata) is not sanitized or validated before being interpolated into the prompt.
  • [EXTERNAL_DOWNLOADS]: The skill's documentation provides instructions to download the core trading runtime and strategy producer scripts directly from the vendor's official GitHub repository.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 28, 2026, 08:48 PM
Security Audit — agent-trust-hub — cub-strategy