compare-literature

Comparing task results against published literature requires gathering task context, extracting published baselines from research files, building a comparison table, and writing a structured report to results/compare_literature.md.

Validating research data with literature benchmarks follows a defined Phase 1-5 workflow: collect context, identify literature benchmarks, build a comparison table, write the report, and verify with the verificator to quantify improvements or gaps.

Yes, the skill reads metrics.json, results_detailed.md, results_summary.md, and research files from per-task directories under tasks/$TASK_ID to identify published results and generate a structured comparison report.

No, the literature comparison workflow has no external dependencies. It operates by reading internal task directories and research files to synthesize published results and construct a standalone verification report.

The best way to document AI task metrics against research baselines is to generate a standardized report file with a structured comparison table and narrative, enabling reproducible validation of autonomous task findings.

Use a systematic literature comparison for task verification when researchers need to validate autonomous task findings by contrasting computed metrics with published baselines to quantify improvement or identify gaps.