Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →用于工单/对话诉求抽取的客户诉求抽取原子 skill，适用于通用行业信息抽取场景。
.claude/skills/aifinlab-customer-intent-extraction/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 91% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 7% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 66% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 108% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 186% | 0% |
本 Skill 支持多种客户交互数据输入格式，核心数据来源包括：
> 说明：本 Skill 不包含数据采集功能，需要用户提供清洗后的对话和工单数据。建议数据包含完整的对话上下文，以便进行准确的诉求抽取。
本 Skill 提供全面的客户诉求抽取能力，涵盖多种抽取功能：
json{ "source_info": { "source_type": "customer_service", "source_id": "CS20240315001", "customer_id": "CUST001", "channel": "online_chat", "timestamp": "2024-03-15T10:30:00" }, "intent_extraction": { "primary_intent": "complaint", "intent_category": "service_quality", "intent_description": "客户投诉服务响应速度慢，要求改进", "intent_keywords": ["响应慢", "服务", "改进"], "intent_confidence": 0.92 }, "content": { "original_text": "你们的服务响应太慢了，我已经等了很久，希望能改进一下。", "extracted_text": "服务响应速度慢，要求改进", "entities": { "customer_name": null, "product_name": null, "time_reference": "很久" } }, "sentiment": { "sentiment": "negative", "sentiment_score": -0.75, "emotion": "frustrated", "intensity": "high" }, "priority": { "urgency": "medium", "importance": "high", "priority_level": 2, "suggested_response_time": "24小时内" }, "related_intents": [ { "intent_id": "INT001", "similarity": 0.85, "description": "服务响应速度相关诉求" } ] }
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 11,737 | 29,477 | +151% | 1 | 1 | 0% | 2,045 | 3,897 | +91% | 0 | 0 | — |
case-02 | fail→fail | 14,564 | 19,762 | +36% | 1 | 1 | 0% | 2,319 | 5,201 | +124% | 0 | 0 | — |
case-03 | fail→fail | 9,914 | 13,672 | +38% | 1 | 1 | 0% | 1,887 | 4,194 | +122% | 0 | 0 | — |
case-04 | fail→fail | 10,327 | 10,752 | +4% | 1 | 1 | 0% | 1,529 | 3,020 | +98% | 0 | 0 | — |
case-05 | fail→pass | 17,732 | 10,889 | -39% | 1 | 1 | 0% | 3,056 | 3,265 | +7% | 0 | 0 | — |
case-06 | fail→pass | 8,951 | 7,545 | -16% | 1 | 1 | 0% | 1,640 | 2,718 | +66% | 0 | 0 | — |
case-07 | fail→pass | 7,916 | 8,219 | +4% | 1 | 1 | 0% | 1,527 | 3,174 | +108% | 0 | 0 | — |
case-08 | fail→pass | 6,980 | 9,665 | +38% | 1 | 1 | 0% | 1,117 | 3,190 | +186% | 0 | 0 | — |
case-09 | fail→pass | 10,427 | 11,683 | +12% | 1 | 1 | 0% | 1,895 | 3,870 | +104% | 0 | 0 | — |
case-10 | fail→fail | 8,384 | 10,075 | +20% | 1 | 1 | 0% | 1,433 | 3,456 | +141% | 0 | 0 | — |
case-11 | pass→pass | 15,992 | 15,215 | -5% | 1 | 1 | 0% | 2,243 | 4,154 | +85% | 0 | 0 | — |
case-12 | fail→pass | 7,646 | 7,746 | +1% | 1 | 1 | 0% | 1,408 | 3,110 | +121% | 0 | 0 | — |
case-13 | fail→pass | 9,337 | 9,364 | +0% | 1 | 1 | 0% | 1,577 | 3,378 | +114% | 0 | 0 | — |
case-14 | pass→pass | 12,016 | 11,671 | -3% | 1 | 1 | 0% | 1,890 | 3,614 | +91% | 0 | 0 | — |
case-15 | fail→pass | 10,770 | 11,712 | +9% | 1 | 1 | 0% | 1,692 | 3,884 | +130% | 0 | 0 | — |
case-16 | fail→pass | 6,407 | 12,152 | +90% | 1 | 1 | 0% | 1,045 | 3,650 | +249% | 0 | 0 | — |
case-17 | fail→fail | 9,048 | 9,688 | +7% | 1 | 1 | 0% | 1,473 | 3,411 | +132% | 0 | 0 | — |
case-18 | fail→pass | 8,806 | 9,645 | +10% | 1 | 1 | 0% | 1,389 | 3,281 | +136% | 0 | 0 | — |
case-19 | pass→pass | 9,447 | 9,813 | +4% | 1 | 1 | 0% | 1,558 | 3,432 | +120% | 0 | 0 | — |
case-20 | fail→fail | 38,324 | 11,204 | -71% | 1 | 1 | 0% | 1,285 | 3,421 | +166% | 0 | 0 | — |
case-21 | fail→pass | 8,235 | 10,057 | +22% | 1 | 1 | 0% | 1,462 | 3,358 | +130% | 0 | 0 | — |
case-22 | fail→pass | 6,085 | 10,694 | +76% | 1 | 1 | 0% | 1,156 | 3,674 | +218% | 0 | 0 | — |
case-23 | pass→pass | 9,786 | 12,118 | +24% | 1 | 1 | 0% | 1,572 | 3,732 | +137% | 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. 23 cases were attempted. The headline lift of +57 percentage points is the difference between those two pass rates over the 23 comparable cases.
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.