data-mining-ai-trading
Pass
Audited by Gen Agent Trust Hub on Apr 25, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill demonstrates safe software engineering practices for data validation. All operations are performed locally using provided scripts.
- [PROMPT_INJECTION]: The script scripts/data_mining_ai_trading_diagnostics.py ingests untrusted data from an external CSV file. It extracts column names and includes them in the output JSON artifact. While this technically creates a surface for indirect prompt injection via malicious column names, the script's focus on numeric transformation significantly limits the potential for exploitation. Ingestion points: scripts/data_mining_ai_trading_diagnostics.py (reads input.csv at line 64). Boundary markers: Absent; instructions do not specify markers for the agent when reading the diagnostic JSON artifact. Capability inventory: The skill uses a Python script for file processing and expects the agent to manage artifacts. No high-risk tools like network access or arbitrary command execution are explicitly enabled. Sanitization: Values are sanitized via numeric conversion (pd.to_numeric), but column names are reflected directly in the output dictionary keys.
Audit Metadata