The Agent Skills Directory

[SAFE]: The skill instructions and referenced methodology files focus exclusively on systematic trading validation. The behavioral patterns observed are consistent with the stated educational and analytical purpose.
[COMMAND_EXECUTION]: The skill includes scripts/evaluate_backtest.py, a script intended to be executed by the agent or user. The script performs statistical calculations on user-supplied metrics and generates reports on the local file system. It does not perform network operations or access sensitive system configuration files.
[INDIRECT_PROMPT_INJECTION]: The skill provides a surface for processing external trading performance data through its evaluation script.
Ingestion points: Numeric metrics (win rate, trade count, drawdown, etc.) are ingested as command-line arguments in scripts/evaluate_backtest.py.
Boundary markers: The script uses argparse with strict type enforcement (integers and floats), which naturally prevents the injection of natural language instructions into the script's logic.
Capability inventory: The script includes file system write capabilities to save JSON and Markdown reports to a user-specified directory.
Sanitization: The validate_inputs function ensures all ingested metrics are within logical ranges (e.g., win rates between 0 and 100, non-negative trade counts), mitigating accidental or intentional data corruption.

backtest-expert