菜单
Use when: running a controlled trial and comparing generated outputs against a baseline folder to classify exact match vs structural equivalence.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use when: extracting reimbursement-relevant fields from receipts, statements, tickets, bookings, and related documents with confidence scoring.
Use when: matching expense evidence to payment proofs and generating explainable confidence-scored reconciliation results.
Use when: users provide a fully noisy unsorted evidence folder and you must inventory, type-triage, and de-duplicate files before extraction.
Use when: running the full ELKE reimbursement pipeline from noisy input to submission-ready output, with sequential baseline and optional parallel subagents.
Use when: assembling the final reimbursement package, enforcing deterministic structure, and producing reviewer-ready quality gate reports.
Use when: performing independent reviewer checks on all generated artifacts and issuing a PASS/FAIL gate decision with blockers.
| name | elke-baseline-comparator |
| description | Use when: running a controlled trial and comparing generated outputs against a baseline folder to classify exact match vs structural equivalence. |
Measure whether a run reproduces baseline outputs exactly or only structurally.
skills/system-prompts/comparator-subagent-system.mdskills/elke-baseline-comparator/templates/baseline_comparison_report.schema.jsonskills/elke-baseline-comparator/templates/comparator_checklist.mdbaseline_rootoutput_rootartifacts_rootWrite to artifacts_root/03_verification/:
baseline_comparison_report.jsonbaseline_comparison_report.mdbaseline_comparison_report.json MUST conform to baseline_comparison_report.schema.json.
exact_samestructurally_equivalent_but_content_divergentdivergent