Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →【深度调研专用】当需要系统性调研、多产品对比、深度分析、技术选型、趋势分析时必须使用本技能。触发关键词：调研、对比、分析、评估、选型、趋势、综合报告、多源验证。通过 3 层深度搜索（阅读多个网页）+ 多源交叉验证，提供比普通 WebSearch 更全面、更准确的研究结果和深度洞察。禁止用普通搜索工具替代深度调研任务。
.claude/skills/aiskillstore-deep-search-and-insight-synthesize/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✓→✗ | ▼ Worse | -21% | 0% |
| case-08 | ✓→✓ | ＝ Same ✓ | 96% | 0% |
| case-09 | ✓→✓ | ＝ Same ✓ | 219% | 0% |
| case-01 | ✗→✗ | ＝ Same ✗ | -60% | 0% |
| case-02 | ✗→✗ | ＝ Same ✗ | -50% | 0% |
本技能提供深度搜索调研能力，专门进行多层级的网络搜索和综合分析，提供类似Perplexity的深度研究功能。核心功能是通过系统性的深度搜索，收集、验证和分析多个信息源，最终提供经过深度思考的综合性见解。
重要：所有的回答和交流都必须使用中文，专业术语名词除外。
执行本技能时应严格遵循以下步骤：
Write工具创建专业的.md文档输出格式要求： 根据问题复杂度，选择合适的输出方式：
行为准则与项目规则：
核心搜索策略：
搜索工具使用：
专业工具集成指导：
输出模式决策逻辑：
安全与限制：
本技能通过系统性的多层级搜索和综合分析，将碎片化的网络信息转化为有价值的洞察和知识。
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 37,016 | 11,819 | -68% | 1 | 1 | 0% | 6,241 | 2,504 | -60% | 0 | 0 | — |
case-02 | fail→fail | 30,043 | 8,868 | -70% | 1 | 1 | 0% | 4,625 | 2,298 | -50% | 0 | 0 | — |
case-03 | fail→fail | 20,138 | 6,728 | -67% | 1 | 1 | 0% | 2,923 | 2,263 | -23% | 0 | 0 | — |
case-04 | fail→fail | 10,701 | 6,759 | -37% | 1 | 1 | 0% | 1,820 | 2,196 | +21% | 0 | 0 | — |
case-05 | fail→fail | 29,250 | 9,652 | -67% | 1 | 1 | 0% | 4,405 | 2,313 | -47% | 0 | 0 | — |
case-06 | fail→fail | 26,927 | 9,341 | -65% | 1 | 1 | 0% | 4,033 | 2,394 | -41% | 0 | 0 | — |
case-07 | pass→fail | 19,735 | 9,451 | -52% | 1 | 1 | 0% | 3,027 | 2,389 | -21% | 0 | 0 | — |
case-08 | pass→pass | 8,759 | 9,562 | +9% | 1 | 1 | 0% | 1,836 | 3,607 | +96% | 0 | 0 | — |
case-09 | pass→pass | 6,111 | 3,663 | -40% | 1 | 1 | 0% | 808 | 2,579 | +219% | 0 | 0 | — |
case-10 | fail→fail | 28,534 | 10,364 | -64% | 1 | 1 | 0% | 4,885 | 2,564 | -48% | 0 | 0 | — |
case-11 | fail→fail | 26,257 | 8,909 | -66% | 1 | 1 | 0% | 4,318 | 2,368 | -45% | 0 | 0 | — |
case-12 | fail→fail | 12,474 | 7,724 | -38% | 1 | 1 | 0% | 2,126 | 2,183 | +3% | 0 | 0 | — |
case-13 | fail→fail | 24,608 | 41,085 | +67% | 1 | 1 | 0% | 4,427 | 8,018 | +81% | 0 | 0 | — |
case-14 | fail→fail | 32,609 | 13,610 | -58% | 1 | 1 | 0% | 6,190 | 2,656 | -57% | 0 | 0 | — |
case-15 | fail→fail | 11,673 | 7,601 | -35% | 1 | 1 | 0% | 1,853 | 2,160 | +17% | 0 | 0 | — |
case-16 | fail→fail | 23,587 | 10,043 | -57% | 1 | 1 | 0% | 4,004 | 2,463 | -38% | 0 | 0 | — |
case-17 | fail→fail | 32,156 | 10,381 | -68% | 1 | 1 | 0% | 4,768 | 2,315 | -51% | 0 | 0 | — |
case-18 | fail→fail | 30,002 | 9,784 | -67% | 1 | 1 | 0% | 6,194 | 2,484 | -60% | 0 | 0 | — |
case-19 | fail→fail | 9,592 | 6,772 | -29% | 1 | 1 | 0% | 1,650 | 2,194 | +33% | 0 | 0 | — |
case-20 | fail→fail | 26,253 | 9,018 | -66% | 1 | 1 | 0% | 4,246 | 2,425 | -43% | 0 | 0 | — |
case-21 | fail→fail | 11,382 | 7,755 | -32% | 1 | 1 | 0% | 1,876 | 2,283 | +22% | 0 | 0 | — |
case-22 | fail→fail | 24,354 | 41,111 | +69% | 1 | 1 | 0% | 4,159 | 2,370 | -43% | 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, and 3 counted toward the lift figure. The other 19 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of -5 percentage points is the difference between those two pass rates over the 3 comparable cases. 11 cases got worse with the skill loaded, and they are included in that figure.
Other measured skills in the registry, with their headline benchmark lift.