/compare

Compares two skills pairwise across quality dimensions, identifying which is better on each.

Tests a skill with parallel with-skill and baseline eval runs across multiple prompts, grades against assertions, and iterates on improvements.

Deep-evaluates a single skill using static analysis and qualitative review, both individually and in context of the full setup. Checks if a skill is worth keeping, well-built, or redundant.

Compares two skill versions or snapshot files, producing a side-by-side dimension score table with delta indicators and trajectory analysis.

Compares two prompt or skill variants, diffs textual changes, scores both on an eval set with paired testing, and reports statistical significance and a ship/don't ship verdict.

Benchmark all skills across the marketplace with static analysis and quality assessment