Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股资本配置/再投资分析/资本支出与股东回报评估。当用户说"资本配置"、"capital allocation"、"回购"、"增发"、"并购"、"钱怎么花的"、"XX在做什么并购"、"定增"、"再融资"、"股份回购"、"资本开支"、"资本配置策略"、"资本配置分析"、"CAPEX"、"股东回报"时触发。MUST USE when user asks about capital allocation strategy, CAPEX analysis, share buyback evaluation, M&A assessment, or how a company deploys its capital. 分析企业资本配置决策（回购/分红/并购/资本开支/偿债），评估管理层的资本配置能力和股东回报意识。支持研报风格（formal）和快速分析风格（brief）。
.claude/skills/aifinlab-a-share-capital-allocation/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-15 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 80% | 0% |
| case-19 | ✓→✗ | ▼ Worse | -43% | 0% |
| case-20 | ✓→✗ | ▼ Worse | -47% | 0% |
| case-21 | ✓→✗ | ▼ Worse | -48% | 0% |
bashSCRIPTS="$SKILLS_ROOT/cn-stock-data/scripts" # 财务数据（利润表/资产负债表/现金流量表 → 资本开支/研发/分红/偿债） python "$SCRIPTS/cn_stock_data.py" finance --code [CODE] # 实时行情（当前市值 → 计算回购/分红收益率） python "$SCRIPTS/cn_stock_data.py" quote --code [CODE] # K线（股价走势 → 回购/增发时点判断） python "$SCRIPTS/cn_stock_data.py" kline --code [CODE] --freq daily --start [3年前] # 资金流向 python "$SCRIPTS/cn_stock_data.py" fund_flow --code [CODE] --days 30
python# 回购数据 import akshare as ak df = ak.stock_repurchase_em() # A股回购汇总
补充 web 搜索：回购公告、定增/配股/可转债方案、并购进展、资本开支计划、分红方案公告。
Step 1: 资本配置全景图（近3-5年）
梳理企业资金去向的完整画面：
| 资金用途 | 占经营现金流比 | 金额(亿) | 趋势 | |---------|-------------|----------|------| | 资本开支(Capex) | | | ↑/→/↓ | | 研发投入 | | | | | 并购(M&A) | | | | | 分红 | | | | | 回购 | | | | | 偿债/降杠杆 | | | | | 账上现金积累 | | | |
数据来源：
Step 2: 资本配置效率评估
再投资回报：
并购评估：
资本结构：
Step 3: 股东回报分析
分红：
回购：
综合股东回报率 = (分红 + 回购注销金额) / 市值
Step 4: 同业对比
| 指标 | 本公司 | 同业A | 同业B | 行业均值 | |------|-------|-------|-------|---------| | ROIC | | | | | | 分红率 | | | | | | 回购/市值 | | | | | | Capex/营收 | | | | | | 商誉/净资产 | | | | | | 股东回报率 | | | | |
Step 5: 输出
| 维度 | formal（机构研报风格） | brief（快速分析风格） | |------|---------------------|---------------------| | 篇幅 | 3-5 页 | 半页-1页 | | 资本配置全景 | 完整表格+3-5年趋势图描述 | 核心数字+方向判断 | | 效率评估 | 详细ROIC/WACC计算、并购逐笔分析 | ROIC vs WACC结论、商誉风险一句话 | | 股东回报 | 分红/回购详细拆解 | 综合回报率+关键判断 | | 同业对比 | 完整对比表 | 1-2个关键指标对比 | | 结论 | 资本配置评级+改进建议 | 一句话评价 | | 免责声明 | 需要 | 不需要 |
默认风格：brief。用户说"详细"、"完整"、"报告"时切换 formal。
python# 调用 skill result = run_skill({ "param1": "value1", "param2": "value2" })
bashpython scripts/run_skill.py --input data.json
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 28,467 | 49,967 | +76% | 1 | 1 | 0% | 4,022 | 1,998 | -50% | 0 | 0 | — |
case-02 | fail→fail | 36,920 | 12,007 | -67% | 1 | 1 | 0% | 5,681 | 2,367 | -58% | 0 | 0 | — |
case-03 | fail→fail | 25,352 | 10,166 | -60% | 1 | 1 | 0% | 3,829 | 2,001 | -48% | 0 | 0 | — |
case-04 | pass→pass | 16,732 | 23,075 | +38% | 1 | 1 | 0% | 2,344 | 4,312 | +84% | 0 | 0 | — |
case-05 | fail→fail | 41,403 | 22,756 | -45% | 1 | 1 | 0% | 3,244 | 4,901 | +51% | 0 | 0 | — |
case-19 | pass→fail | 42,950 | 61,992 | +44% | 1 | 1 | 0% | 3,679 | 2,100 | -43% | 0 | 0 | — |
case-20 | pass→fail | 28,990 | 36,643 | +26% | 1 | 1 | 0% | 3,783 | 1,997 | -47% | 0 | 0 | — |
case-21 | pass→fail | 24,267 | 44,391 | +83% | 1 | 1 | 0% | 3,684 | 1,924 | -48% | 0 | 0 | — |
case-06 | pass→pass | 19,801 | 18,878 | -5% | 1 | 1 | 0% | 2,660 | 3,969 | +49% | 0 | 0 | — |
case-07 | pass→pass | 13,307 | 9,764 | -27% | 1 | 1 | 0% | 1,983 | 3,178 | +60% | 0 | 0 | — |
case-08 | pass→pass | 24,079 | 21,152 | -12% | 1 | 1 | 0% | 3,228 | 4,540 | +41% | 0 | 0 | — |
case-09 | fail→fail | 35,865 | 13,633 | -62% | 1 | 1 | 0% | 4,619 | 2,151 | -53% | 0 | 0 | — |
case-10 | fail→fail | 17,501 | 10,630 | -39% | 1 | 1 | 0% | 2,175 | 1,943 | -11% | 0 | 0 | — |
case-11 | pass→pass | 27,422 | 34,465 | +26% | 1 | 1 | 0% | 3,465 | 5,480 | +58% | 0 | 0 | — |
case-12 | pass→pass | 25,015 | 16,299 | -35% | 1 | 1 | 0% | 2,970 | 3,809 | +28% | 0 | 0 | — |
case-13 | pass→pass | 13,871 | 9,273 | -33% | 1 | 1 | 0% | 1,840 | 2,864 | +56% | 0 | 0 | — |
case-14 | pass→pass | 25,610 | 23,619 | -8% | 1 | 1 | 0% | 3,568 | 5,177 | +45% | 0 | 0 | — |
case-15 | fail→pass | 24,255 | 14,680 | -39% | 1 | 1 | 0% | 3,160 | 4,023 | +27% | 0 | 0 | — |
case-16 | pass→pass | 20,877 | 24,036 | +15% | 1 | 1 | 0% | 3,191 | 4,425 | +39% | 0 | 0 | — |
case-17 | fail→pass | 17,993 | 19,830 | +10% | 1 | 1 | 0% | 2,233 | 4,023 | +80% | 0 | 0 | — |
case-18 | pass→pass | 18,487 | 51,525 | +179% | 1 | 1 | 0% | 2,421 | 4,672 | +93% | 0 | 0 | — |
case-22 | pass→pass | 44,960 | 29,148 | -35% | 1 | 1 | 0% | 3,394 | 4,920 | +45% | 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 14 counted toward the lift figure. The other 8 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 14 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.