Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →元典案例检索技能（开放平台版 https://open.chineselaw.com）。 用于中国案例、类案、典型案例、权威案例的关键词检索与语义检索，以及案例详情拉取。 不负责最终法律结论；如需基于案例做论证，应继续调用 cs-china-lawyer-analyst； 若是"做法律研究备忘录"类完整流程，应交由 legal-research 编排技能调度。
.claude/skills/thomasmoreai-yd-case-search/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 18% | 0% |
| case-20 | ✗→✓ | ▲ Improved | 77% | 0% |
| case-09 | ✓→✗ | ▼ Worse | -60% | 0% |
| case-15 | ✓→✓ | ＝ Same ✓ | -20% | 0% |
| case-22 | ✓→✓ | ＝ Same ✓ | 27% | 0% |
案例检索型 skill，负责把相关案例和类案快速捞出来。
yd-case-search：找案例（本 skill）yd-law-search：找法条和规范依据cs-china-lawyer-analyst：把案例和法条整合成法律意见legal-research：上层编排技能，按"查法规→查案例→出备忘录"调度上述三者与 yd-law-search 相同，按以下顺序读取：
YD_OPEN_API_KEY~/.yd_open_api_keyYD_API_KEY~/.yd_api_key.api_key申请入口：<https://open.chineselaw.com/>
case_detail 配合 id 与 type（ptal/qwal）拉取。--full 开关用于调试或用户明确要求全文展示时使用。search_al — 案例关键词/筛选检索适合：已有关键词、案由、法院、案号、标题，要做筛选式案例研究。
路由：
authority_only=true → 走 /open/rh_qwal_search（权威/典型案例）/open/rh_ptal_search（普通案例）search_al_vector — 案例语义检索适合：用自然语言描述争议焦点、做类案语义检索。
case_detail — 案例详情按 id 拉单个案例全文。必须传 type：ptal（普通）或 qwal（权威）。
bash# 1. 关键词检索（普通案例） echo '{"ay":["买卖合同纠纷"],"xzqh_p":["北京"],"qw":"违约金 逾期","top_k":5}' | \ python3 ~/.claude/skills/yd-case-search/scripts/yd_case_search.py search_al --stdin # 2. 关键词检索（权威案例） echo '{"qw":"人工智能","authority_only":true}' | \ python3 ~/.claude/skills/yd-case-search/scripts/yd_case_search.py search_al --stdin # 3. 案例语义检索（带筛选） echo '{"query":"正当防卫的限度","wenshu_filter":{"ja_start":"2020-01-01","ja_end":"2025-12-31"}}' | \ python3 ~/.claude/skills/yd-case-search/scripts/yd_case_search.py search_al_vector --stdin # 4. 案例详情（按 id） echo '{"id":"183fe9bf6e95f51fab804e383854a51f","type":"ptal"}' | \ python3 ~/.claude/skills/yd-case-search/scripts/yd_case_search.py case_detail --stdin
通用 flag：
--stdin / --full / --raw（同 yd-law-search）search_al_vectorsearch_alauthority_only=truetop_k = 8；展示时至少保留：标题、案号、案件类别、经办法院、裁判日期、id（含 type）、链接cs-china-lawyer-analyst| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 20,362 | 5,803 | -72% | 1 | 1 | 0% | 3,633 | 1,571 | -57% | 0 | 0 | — |
case-02 | fail→fail | 17,396 | 6,365 | -63% | 1 | 1 | 0% | 3,372 | 1,607 | -52% | 0 | 0 | — |
case-03 | fail→fail | 26,700 | 6,279 | -76% | 1 | 1 | 0% | 4,714 | 1,556 | -67% | 0 | 0 | — |
case-04 | fail→fail | 22,081 | 7,140 | -68% | 1 | 1 | 0% | 3,769 | 1,615 | -57% | 0 | 0 | — |
case-05 | fail→fail | 16,839 | 6,418 | -62% | 1 | 1 | 0% | 2,996 | 1,645 | -45% | 0 | 0 | — |
case-06 | fail→fail | 5,237 | 4,705 | -10% | 1 | 1 | 0% | 954 | 1,533 | +61% | 0 | 0 | — |
case-07 | fail→pass | 6,491 | 4,100 | -37% | 1 | 1 | 0% | 1,160 | 1,370 | +18% | 0 | 0 | — |
case-08 | fail→fail | 18,518 | 6,637 | -64% | 1 | 1 | 0% | 3,123 | 1,591 | -49% | 0 | 0 | — |
case-09 | pass→fail | 22,862 | 7,512 | -67% | 1 | 1 | 0% | 3,889 | 1,538 | -60% | 0 | 0 | — |
case-10 | fail→fail | 18,879 | 8,217 | -56% | 1 | 1 | 0% | 3,148 | 1,665 | -47% | 0 | 0 | — |
case-11 | fail→fail | 20,206 | 6,861 | -66% | 1 | 1 | 0% | 3,619 | 1,677 | -54% | 0 | 0 | — |
case-12 | fail→fail | 14,386 | 5,539 | -61% | 1 | 1 | 0% | 2,766 | 1,497 | -46% | 0 | 0 | — |
case-13 | fail→fail | 21,876 | 6,020 | -72% | 1 | 1 | 0% | 4,207 | 1,417 | -66% | 0 | 0 | — |
case-14 | fail→fail | 13,669 | 5,480 | -60% | 1 | 1 | 0% | 2,155 | 1,487 | -31% | 0 | 0 | — |
case-15 | pass→pass | 8,966 | 2,178 | -76% | 1 | 1 | 0% | 1,871 | 1,493 | -20% | 0 | 0 | — |
case-16 | fail→fail | 19,820 | 5,848 | -70% | 1 | 1 | 0% | 3,205 | 1,403 | -56% | 0 | 0 | — |
case-17 | fail→fail | 17,605 | 6,238 | -65% | 1 | 1 | 0% | 3,035 | 1,548 | -49% | 0 | 0 | — |
case-18 | fail→fail | 17,087 | 5,586 | -67% | 1 | 1 | 0% | 2,970 | 1,580 | -47% | 0 | 0 | — |
case-19 | fail→fail | 17,105 | 6,463 | -62% | 1 | 1 | 0% | 3,656 | 1,659 | -55% | 0 | 0 | — |
case-20 | fail→pass | 13,307 | 17,318 | +30% | 1 | 1 | 0% | 2,653 | 4,689 | +77% | 0 | 0 | — |
case-21 | fail→fail | 23,856 | 6,506 | -73% | 1 | 1 | 0% | 4,371 | 1,614 | -63% | 0 | 0 | — |
case-22 | pass→pass | 18,764 | 16,671 | -11% | 1 | 1 | 0% | 3,155 | 3,997 | +27% | 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 4 counted toward the lift figure. The other 18 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 4 comparable cases. 4 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.