The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local Python script scripts/backtesting_validation_diagnostics.py to process CSV data and generate diagnostic artifacts. This is consistent with its stated purpose of quantitative research validation.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided CSV data via the pandas library. While it ingests external data, the capabilities are restricted to calculating statistical summaries (mean, standard deviation, correlation, drawdown) and writing results to a local JSON file. No executable code or instructions are extracted from the data into the agent's prompt context.
[DATA_EXPOSURE]: The skill reads and writes local files (CSV and JSON). There are no network operations or calls to external domains, preventing unauthorized data transmission.

backtesting-validation