Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →証拠優先のECC現状調査ワークフロー。ユーザーが現在の公開証拠と提供されたローカルコンテキストに基づいて最新の事実、比較、情報の充実、または推奨事項を求める場合に使用する。
.claude/skills/affaan-m-research-ops/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 14% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 47% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 6% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 13% | 0% |
当用户要求研究当前信息、比较选项、丰富人员或公司信息，或将重复查询转化为可监控的工作流时，使用此功能。
这是仓库研究栈的操作封装。它并非 deep-research、exa-search 或 market-research 的替代品；而是指示何时以及如何将它们结合使用。
在相关场景下，将这些 ECC 原生技能纳入工作流：
exa-search：用于快速发现当前网络信息deep-research：用于多源综合并附带引用market-research：当最终结果应为建议或排序决策时使用lead-intelligence：当任务针对人员/公司而非通用研究时使用knowledge-ops：当结果需持久存储于后续上下文时使用将任何提供的材料规范化为：
如果用户已构建部分模型，不要从零开始重新分析。
在搜索前选择正确的路径：
exa-search 进行快速发现deep-researchmarket-researchlead-intelligence对于重要声明，说明其属于：
对时效性敏感的答案应包含具体日期。
如果用户可能反复提出相同的研究问题，请明确说明，并建议采用监控或工作流层，而非永远重复相同的手动搜索。
text问题类型 - 事实性 / 比较性 / 补充性 / 监控性 证据 - 有来源的事实 - 用户提供的上下文 推论 - 从证据中得出的结论 建议 - 答案或下一步行动 - 是否应将其设为监控项
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 24,653 | 22,963 | -7% | 1 | 1 | 0% | 3,672 | 4,272 | +16% | 0 | 0 | — |
case-02 | fail→fail | 23,922 | 22,638 | -5% | 1 | 1 | 0% | 3,573 | 4,263 | +19% | 0 | 0 | — |
case-03 | fail→fail | 22,861 | 19,690 | -14% | 1 | 1 | 0% | 3,345 | 3,768 | +13% | 0 | 0 | — |
case-04 | pass→fail | 13,502 | 12,885 | -5% | 1 | 1 | 0% | 2,194 | 2,865 | +31% | 0 | 0 | — |
case-05 | pass→fail | 7,482 | 6,835 | -9% | 1 | 1 | 0% | 1,015 | 1,945 | +92% | 0 | 0 | — |
case-06 | pass→pass | 20,602 | 19,681 | -4% | 1 | 1 | 0% | 2,377 | 2,950 | +24% | 0 | 0 | — |
case-07 | fail→pass | 13,161 | 10,183 | -23% | 1 | 1 | 0% | 2,234 | 2,550 | +14% | 0 | 0 | — |
case-08 | fail→pass | 23,363 | 35,819 | +53% | 1 | 1 | 0% | 3,555 | 4,279 | +20% | 0 | 0 | — |
case-09 | fail→pass | 21,646 | 21,122 | -2% | 1 | 1 | 0% | 2,899 | 4,263 | +47% | 0 | 0 | — |
case-10 | fail→pass | 22,561 | 18,491 | -18% | 1 | 1 | 0% | 3,646 | 3,878 | +6% | 0 | 0 | — |
case-11 | fail→pass | 19,503 | 14,813 | -24% | 1 | 1 | 0% | 2,926 | 3,307 | +13% | 0 | 0 | — |
case-12 | fail→pass | 20,955 | 24,057 | +15% | 1 | 1 | 0% | 3,369 | 4,294 | +27% | 0 | 0 | — |
case-13 | fail→pass | 22,443 | 8,899 | -60% | 1 | 1 | 0% | 3,645 | 2,221 | -39% | 0 | 0 | — |
case-14 | fail→pass | 18,996 | 21,261 | +12% | 1 | 1 | 0% | 3,097 | 4,295 | +39% | 0 | 0 | — |
case-15 | fail→pass | 16,409 | 16,916 | +3% | 1 | 1 | 0% | 2,476 | 3,437 | +39% | 0 | 0 | — |
case-16 | fail→pass | 19,544 | 17,321 | -11% | 1 | 1 | 0% | 3,221 | 3,726 | +16% | 0 | 0 | — |
case-17 | fail→pass | 22,453 | 17,277 | -23% | 1 | 1 | 0% | 3,212 | 3,400 | +6% | 0 | 0 | — |
case-18 | fail→pass | 17,522 | 15,809 | -10% | 1 | 1 | 0% | 2,743 | 3,038 | +11% | 0 | 0 | — |
case-19 | fail→pass | 22,349 | 16,272 | -27% | 1 | 1 | 0% | 3,227 | 3,159 | -2% | 0 | 0 | — |
case-20 | fail→pass | 17,076 | 21,845 | +28% | 1 | 1 | 0% | 2,522 | 4,061 | +61% | 0 | 0 | — |
case-21 | fail→pass | 22,768 | 14,875 | -35% | 1 | 1 | 0% | 3,246 | 3,076 | -5% | 0 | 0 | — |
case-22 | fail→pass | 46,662 | 25,863 | -45% | 1 | 1 | 0% | 2,924 | 3,814 | +30% | 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 +64 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.