defi-trading-systems

Pass

Audited by Gen Agent Trust Hub on Mar 9, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted financial data from external sources to drive trading logic, risk calculations, and MEV protection strategies. \n
  • Ingestion points: Market data is ingested in LiquidationMonitor.check_health (SKILL.md), FundingArbitrage.find_opportunities (SKILL.md), and detect_sandwich (references/mev-protection.md). \n
  • Boundary markers: There are no explicit delimiters or system instructions to ignore embedded instructions within the processed financial data. \n
  • Capability inventory: The skill has network capabilities via requests.post for submitting transaction bundles to Flashbots relays (SKILL.md and references/mev-protection.md). \n
  • Sanitization: No validation or sanitization logic is present for external price feeds or transaction data before they are used in logical branches.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 9, 2026, 10:10 PM