Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →用于银行财富管理场景下生成合规话术与客户沟通主线，适合在需要输出可直接使用的沟通框架、禁用表述提醒与后续跟进建议时触发。
.claude/skills/aifinlab-bank-t195-wealth-management-compliant-talk-track-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 62% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -28% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 68% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 51% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 67% | 0% |
本技能用于在银行财富管理场景中，把客户目标、风险承受能力、期限与流动性约束、产品适当性与市场观点组织成合规可用的沟通主线与话术内容。适用于理财经理、财富顾问、私行团队在客户触达、路演、陪伴、波动沟通与持仓解释等环节的合规表达。
当需要批量生成/检查话术，或需要稳定输出结构化内容时，使用 scripts/ 中的脚本：
scripts/talk_track_builder.py：读取结构化输入，生成合规话术主线、话术片段与风险提示。scripts/phrase_guard.py：对话术进行禁用表述扫描，标注替换建议与风险等级。talk_track_builder.py 生成初稿phrase_guard.py 扫描禁用表述与高风险措辞| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 22,912 | 37,756 | +65% | 1 | 1 | 0% | 3,318 | 4,464 | +35% | 0 | 0 | — |
case-02 | fail→fail | 33,491 | 28,088 | -16% | 1 | 1 | 0% | 4,567 | 5,123 | +12% | 0 | 0 | — |
case-03 | fail→fail | 24,758 | 25,375 | +2% | 1 | 1 | 0% | 3,659 | 4,792 | +31% | 0 | 0 | — |
case-04 | pass→pass | 14,119 | 10,172 | -28% | 1 | 1 | 0% | 2,092 | 2,110 | +1% | 0 | 0 | — |
case-05 | pass→pass | 19,114 | 19,431 | +2% | 1 | 1 | 0% | 2,319 | 3,985 | +72% | 0 | 0 | — |
case-06 | pass→pass | 13,195 | 13,294 | +1% | 1 | 1 | 0% | 1,983 | 2,871 | +45% | 0 | 0 | — |
case-07 | fail→pass | 21,918 | 27,502 | +25% | 1 | 1 | 0% | 2,871 | 4,651 | +62% | 0 | 0 | — |
case-08 | fail→fail | 19,928 | 23,297 | +17% | 1 | 1 | 0% | 2,679 | 4,116 | +54% | 0 | 0 | — |
case-09 | fail→pass | 22,900 | 8,447 | -63% | 1 | 1 | 0% | 3,036 | 2,198 | -28% | 0 | 0 | — |
case-10 | fail→fail | 23,208 | 26,635 | +15% | 1 | 1 | 0% | 2,884 | 4,325 | +50% | 0 | 0 | — |
case-11 | fail→pass | 23,090 | 32,648 | +41% | 1 | 1 | 0% | 2,945 | 4,944 | +68% | 0 | 0 | — |
case-12 | fail→fail | 23,742 | 26,193 | +10% | 1 | 1 | 0% | 2,794 | 4,569 | +64% | 0 | 0 | — |
case-13 | pass→fail | 20,399 | 23,480 | +15% | 1 | 1 | 0% | 2,530 | 3,852 | +52% | 0 | 0 | — |
case-14 | fail→fail | 20,560 | 25,094 | +22% | 1 | 1 | 0% | 2,519 | 4,324 | +72% | 0 | 0 | — |
case-15 | fail→pass | 21,043 | 25,580 | +22% | 1 | 1 | 0% | 2,873 | 4,352 | +51% | 0 | 0 | — |
case-16 | fail→pass | 17,353 | 26,025 | +50% | 1 | 1 | 0% | 2,514 | 4,198 | +67% | 0 | 0 | — |
case-17 | fail→pass | 25,361 | 26,671 | +5% | 1 | 1 | 0% | 2,901 | 4,299 | +48% | 0 | 0 | — |
case-18 | fail→fail | 17,975 | 25,472 | +42% | 1 | 1 | 0% | 2,506 | 4,575 | +83% | 0 | 0 | — |
case-19 | fail→pass | 18,060 | 24,783 | +37% | 1 | 1 | 0% | 2,780 | 3,957 | +42% | 0 | 0 | — |
case-20 | fail→fail | 32,897 | 34,084 | +4% | 1 | 1 | 0% | 2,948 | 4,442 | +51% | 0 | 0 | — |
case-21 | pass→pass | 21,017 | 28,450 | +35% | 1 | 1 | 0% | 2,615 | 4,304 | +65% | 0 | 0 | — |
case-22 | pass→pass | 24,302 | 30,756 | +27% | 1 | 1 | 0% | 3,015 | 4,395 | +46% | 0 | 0 | — |
case-23 | fail→pass | 18,620 | 9,328 | -50% | 1 | 1 | 0% | 2,511 | 2,150 | -14% | 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 +30 percentage points is the difference between those two pass rates over the 23 comparable cases. 1 case got worse with the skill loaded, and it is 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.