ml4t-run-backtest
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFE
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill references the
ml4t-backtestlibrary in its metadata and imports theml4t.backtestmodule in code examples. These are vendor-owned resources associated with the author 'ml4t' and represent legitimate functionality for the stated purpose of trading simulation. No unauthorized or suspicious remote code execution patterns were found. - [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: The skill defines an interface for processing external data frames (
prices_df,signals_df) via theDataFeedclass inSKILL.md. - Boundary markers: None present for the input data structures.
- Capability inventory: The skill's logic is restricted to mathematical simulations using
numpyand the vendor's backtesting engine. There are no subprocess calls, network operations, file-write capabilities, or dynamic execution (eval/exec) functions present. - Sanitization: No sanitization is performed, but no dangerous capabilities exist to exploit.
- [SAFE]: The skill was inspected for prompt injection, obfuscation, credential exposure, and persistence mechanisms. No malicious patterns or security vulnerabilities were identified. All code examples use standard Python libraries for financial modeling.
Audit Metadata