Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →ECCの証拠優先の収益、価格設定、返金、チーム請求、請求モデルの実態確認ワークフロー。ユーザーが販売スナップショット、価格比較、重複請求の診断、または汎用的な支払いアドバイスではなくコードに裏付けられた請求の実態を必要とする場合に使用します。
.claude/skills/affaan-m-finance-billing-ops/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -14% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 10% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 54% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 231% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -6% | 0% |
当用户想要了解资金、定价、退款、团队席位逻辑，或产品是否真的如网站和销售文案所暗示的那样运作时，使用此技能。
此技能比 customer-billing-ops 更广泛。该技能用于客户补救。此技能用于运营者真相：收入状态、定价决策、团队计费以及基于代码的计费行为。
在相关时，将这些 ECC 原生技能引入工作流程：
customer-billing-ops 用于特定客户的补救和跟进research-ops 当竞争对手定价或当前市场证据重要时market-research 当答案应以定价建议结束时github-ops 当计费真相取决于兄弟仓库中的代码、待办事项或发布状态时verification-loop 当答案取决于验证结账、席位处理或权限行为时优先使用实时计费数据。如果数据不是实时的，请明确说明快照时间戳。
规范化视图：
如果问题是针对特定客户的，请先分类：
然后将其与更广泛的产品问题分开：
如果答案取决于实现真相，请检查代码路径：
报告：
text快照 - 时间戳 - 收入 / 订阅 / 异常 客户影响 - 谁受影响 - 发生了什么 产品真相 - 代码实际执行的操作 - 网站或销售文案声称的内容 决策 - 退款 / 保留 / 转化 / 无操作 产品差距 - 需要构建或修复的具体后续事项
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 47,771 | 13,284 | -72% | 1 | 1 | 0% | 3,300 | 2,852 | -14% | 0 | 0 | — |
case-02 | fail→pass | 19,406 | 16,311 | -16% | 1 | 1 | 0% | 2,855 | 3,153 | +10% | 0 | 0 | — |
case-21 | pass→pass | 36,672 | 25,500 | -30% | 1 | 1 | 0% | 5,656 | 4,755 | -16% | 0 | 0 | — |
case-03 | fail→pass | 11,167 | 14,722 | +32% | 1 | 1 | 0% | 1,919 | 2,953 | +54% | 0 | 0 | — |
case-04 | pass→pass | 16,967 | 11,756 | -31% | 1 | 1 | 0% | 2,578 | 2,476 | -4% | 0 | 0 | — |
case-05 | fail→pass | 8,935 | 29,424 | +229% | 1 | 1 | 0% | 920 | 3,049 | +231% | 0 | 0 | — |
case-06 | pass→pass | 8,868 | 7,787 | -12% | 1 | 1 | 0% | 1,531 | 2,093 | +37% | 0 | 0 | — |
case-07 | fail→pass | 25,535 | 13,429 | -47% | 1 | 1 | 0% | 3,002 | 2,821 | -6% | 0 | 0 | — |
case-08 | pass→pass | 14,997 | 15,221 | +1% | 1 | 1 | 0% | 2,524 | 3,169 | +26% | 0 | 0 | — |
case-09 | pass→pass | 15,808 | 13,037 | -18% | 1 | 1 | 0% | 2,295 | 2,614 | +14% | 0 | 0 | — |
case-10 | fail→fail | 15,515 | 15,338 | -1% | 1 | 1 | 0% | 2,598 | 2,996 | +15% | 0 | 0 | — |
case-11 | pass→pass | 19,053 | 13,681 | -28% | 1 | 1 | 0% | 3,036 | 3,011 | -1% | 0 | 0 | — |
case-22 | pass→pass | 10,729 | 9,871 | -8% | 1 | 1 | 0% | 1,819 | 2,746 | +51% | 0 | 0 | — |
case-12 | pass→pass | 14,680 | 14,339 | -2% | 1 | 1 | 0% | 2,086 | 2,923 | +40% | 0 | 0 | — |
case-13 | pass→pass | 23,257 | 11,060 | -52% | 1 | 1 | 0% | 2,655 | 2,411 | -9% | 0 | 0 | — |
case-14 | pass→pass | 18,873 | 11,981 | -37% | 1 | 1 | 0% | 2,695 | 2,601 | -3% | 0 | 0 | — |
case-15 | pass→pass | 15,971 | 13,389 | -16% | 1 | 1 | 0% | 2,406 | 2,882 | +20% | 0 | 0 | — |
case-16 | pass→pass | 14,656 | 11,996 | -18% | 1 | 1 | 0% | 2,017 | 2,407 | +19% | 0 | 0 | — |
case-17 | fail→fail | 12,912 | 3,853 | -70% | 1 | 1 | 0% | 1,900 | 1,314 | -31% | 0 | 0 | — |
case-18 | pass→pass | 15,089 | 13,297 | -12% | 1 | 1 | 0% | 2,290 | 2,854 | +25% | 0 | 0 | — |
case-19 | fail→pass | 15,942 | 10,672 | -33% | 1 | 1 | 0% | 2,399 | 2,347 | -2% | 0 | 0 | — |
case-20 | pass→fail | 4,553 | 7,045 | +55% | 1 | 1 | 0% | 843 | 2,065 | +145% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +23 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.