research-harness-work

To generate reproducible analysis scripts for auditability, execute approved tasks from analysis_plan.md to automatically produce scripts with enforced header comments, project-relative paths, seed control, and logged outputs.

The best way to maintain audit logs for research workflow automation is to execute approved analysis tasks in a controlled environment that captures all log files and output artifacts while validating them against an evidence ledger.

Yes, you can run reproducible scripts in Python, R, or Stata, as the execution environment enforces seed control, header comments, and output capture based on the scripting languages defined by your study specifications.

Yes, you need a human-approved analysis plan before executing reproducible research scripts, because the skill operates on tasks defined in analysis_plan.md after planning is complete and before the review phase.

You ensure reproducibility when operating on processed datasets under data/processed/ by enforcing project-relative paths, seed control, and capturing all generated log files and output artifacts during script execution.

A limitation of automated research workflow governance is that it requires integration with separate planning and review stages, meaning it cannot execute tasks lacking prior human approval or validate outputs outside the evidence ledger.