adaptive-wfo-epoch
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides detailed mathematical formulations and code templates for Walk-Forward Optimization (WFO) and Walk-Forward Efficiency (WFE). The implementations use standard, trusted libraries including NumPy, SciPy, Pandas, and PyTorch.
- [DATA_EXFILTRATION]: No network exfiltration or credential harvesting patterns were detected. All network operations mentioned in documentation (such as arXiv links) are standard academic references and do not represent executable code or data leaks.
- [COMMAND_EXECUTION]: While 'Bash' is listed in the allowed-tools configuration, there are no scripts or instructions that execute arbitrary or dangerous shell commands. The usage of shell-based tools is consistent with managing repository files and executing standard Python scripts.
- [PROMPT_INJECTION]: There are no patterns of prompt injection, safety filter bypasses, or instructions to ignore system guidelines. The 'Self-Evolving Skill' instruction is a standard maintenance meta-instruction for autonomous agents to keep documentation updated with reproducible fixes.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests numerical financial data for processing. It includes significant 'guardrails' and 'principled guidelines' to prevent statistical errors like look-ahead bias and overfitting, which indirectly secures the agent's logic against manipulated data distributions.
- [OBFUSCATION]: Analysis of the skill body and all 21 reference files revealed no obfuscated code, Base64 strings, zero-width characters, or homoglyph-based attacks.
Audit Metadata