harvest

To collect SLURM experiment results automatically, the harvest skill detects completed jobs via marker files or sacct scans and centralizes their outputs into docs/experiments. It analyzes both local and remote clusters to generate structured summaries.

Yes, you can update experiment documentation across remote SLURM clusters. The harvest skill analyzes sacct data to cover remote environments and automatically updates docs/experiments.md with concise summaries from all detected jobs.

The best way to track SLURM job metrics is by scanning sacct data and marker files. This process locates finished experiments, collects metrics from logs, and produces structured detail and index updates to keep your records current.

Yes, the harvest skill works with both sacct data and marker files to detect completed SLURM jobs. It analyzes these multiple sources to ensure comprehensive discovery of harvestable experiments across your computing environments.

To generate SLURM job summaries and update your experiments index, run the harvest skill to locate finished jobs. It collects outputs from logs and automatically writes structured detail and index updates to docs/experiments.

When collecting SLURM experiment results, the process enforces safe data handling and logs steps to prevent errors. It relies on detecting marker files or sacct data, so jobs lacking these signals on local or remote clusters may not be discovered.