research-report

Run /research-report in the directory containing your outline.yaml. The skill scans all JSON results, asks which summary fields to show in the table of contents, generates a generate_report.py script, and runs it to produce report.md.

The generated Python script supports both flat JSON and nested category structures, looking up fields at the top level, via category mapping keys, or by traversing nested dicts. Complex values like lists of dicts are formatted line by line for readability.

No. Fields whose values contain the [uncertain] marker, fields listed in the uncertain array, and fields with None or empty values are skipped. The uncertain field names are listed separately in an Other Info section.

Yes. The skill scans the JSON results for short numeric or metric fields such as github_stars or swe_bench_score, then asks via request_user_input which of those fields to display next to each item name in the TOC.

The workflow requires Python with pyyaml installed to read the outline.yaml and fields.yaml configuration files. The generated generate_report.py script reads the JSON results and writes the final report.md.