JADE Skill Matrix Evaluator

Objective

Classify each pipeline skill as draft, experimental, candidate, or official based on run evidence.

Scoring dimensions

• contract compliance (adherence to SKILL.md constraints)
• reproducibility (deterministic outputs across runs)
• gate pass rate (rule pipeline pass %)
• artifact completeness (all expected files present+valid)
• failure handling quality (graceful degradation on missing data)

Classification thresholds

• < 50% aggregate → draft
• 50-69% → experimental
• 70-89% → candidate
• >= 90% → official

Inputs

• artifacts from full run (00-run-config.json, 05-rule-queue.json, rule-status.json, per-rule batch artifacts)

Outputs

• artifacts/10-skill-matrix.json

Use scripts/evaluate_skills.py to execute.