agent-trading-predictor

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The skill utilizes a sandbox tool (mcp__flow-nexus__sandbox_execute) to run dynamically generated Python code for trading operations.\n- [INDIRECT_PROMPT_INJECTION]: The agent is designed to ingest and process market data and trading signals from external sources, which is a common surface for indirect prompt injection attacks.\n
  • Ingestion points: Market data feeds accessed via connect_market_feeds() and external signal vectors like marketSignalVector.\n
  • Boundary markers: No explicit delimiting or validation instructions for external data are provided in the prompt context.\n
  • Capability inventory: The skill has access to automated trading execution tools (predictWithTemporalAdvantage) and a code execution sandbox.\n
  • Sanitization: There are no instructions for sanitizing or filtering input from external financial data sources.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 11:38 AM
Security Audit — agent-trust-hub — agent-trading-predictor