agentsclimarketplace

Octagon sec segment reporting

Skill findscripter/everything-skills/09-verticals/octagon-sec-segment-reporting

当需要从美股公司 10-K/10-Q 文件中研究业务分部(segment)的营收、营业利润、利润率、地区拆分与分部重组时使用;通过 Octagon MCP 的 octagon-agent 工具做分部业绩分析并产出结构化对比表与趋势解读;不适用于无 Octagon MCP 环境、非上市公司或整体财报(非分部维度)分析。触发词:分部业绩、segment revenue、10-K分部From its SKILL.md

Install
npx -y skills add findscripter/everything-skills --skill octagon-sec-segment-reporting

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its file declares

Copied from the file, not written here

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

6.0 KB, ~1.8k tokens by cl100k_base, as published. Nobody here has run it

何时使用

当你需要拆解一家美股上市公司的业务分部表现时使用,典型问题:

  • 各业务分部(segment)的营收、同比增速、营业利润、利润率分别是多少?
  • 公司按地区(Americas/EMEA/APAC)的营收如何分布?
  • 公司近期是否调整/重组了分部口径,怎么调的?
  • 对标两家公司(如 META vs GOOGL)的分部结构与盈利能力。
  • 为估值做 sum-of-parts(分部加总)分析。

不该用的边界:

  • 未在 AI 客户端(Cursor / Claude Desktop / Windsurf 等)配置 Octagon MCP 时——本技能依赖该 MCP,先去配置。
  • 非上市公司、无 SEC 申报的主体——数据源取不到。
  • 只关注公司整体财报(合并口径),不涉及分部维度——直接查财报即可,无需本技能。
  • 需要原始 XBRL 逐行核对或非美股市场——超出范围。

步骤

  1. 确定分析参数

    • Ticker:股票代码(如 AAPL、MSFT、IBM)。
    • 申报类型(可选):10-K(年报)或 10-Q(季报)。
    • 关注点(可选):营收 / 盈利能力 / 地区 / 趋势。
  2. 通过 Octagon MCP 发起查询:调用 octagon-agent 工具,传入自然语言 prompt。

  3. 读取结构化输出:分部对比表 + 关键趋势 + 数据源说明。

  4. 解读结果:理解分部构成、评估盈利能力、追踪重组、分析地区分部(详见下方"注意事项"中的框架)。

指令

确认 Octagon MCP 已配置后,使用 octagon-agent 工具,prompt 模板:

Analyze business segment performance and reporting from <TICKER>'s latest quarterly filing.

MCP 调用格式:

{
  "server": "octagon-mcp",
  "toolName": "octagon-agent",
  "arguments": {
    "prompt": "Analyze business segment performance and reporting from IBM's latest quarterly filing."
  }
}

返回为结构化分部分析,含分部表、关键趋势,数据源标注为 octagon-financials-agentoctagon-sec-agent

示例

完整分部分析:

Analyze business segment performance and reporting from IBM's latest quarterly filing.

分部盈利能力对比:

Compare operating margins across MSFT's business segments in the latest 10-K.

地区分部:

Analyze AAPL's revenue breakdown by geographic region from the latest 10-K.

分部增长趋势:

Track revenue growth trends by segment for GOOGL over the last 4 quarters.

分部重组:

Has AMZN restructured its segment reporting in recent filings and how?

跨公司对标:

Compare segment performance between META and GOOGL in their latest annual filings.

返回示例(IBM 季报):

SegmentRevenueYoY GrowthOperating IncomeMargin
Software$7,209M10.5%$2,374M32.9%
Consulting$5,324M3.3%$686M12.9%
Infrastructure$3,559M17.0%$644M18.1%
Financing$200M10.4%$123M61.6%

注意事项

分部报告框架(ASC 280):营业分部披露营收(含分部间内部销售)、营业利润、分部资产、CapEx、折旧摊销;地区分部披露分地区营收、长期资产、重大国别拆分。

关键指标速算:

  • 营收占比 = 分部营收 / 总营收(看业务构成)。
  • 增速 =(本期 − 上期)/ 上期。
  • 营业利润率 = 营业利润 / 营收(看效率)。
  • ROA = 营业利润 / 分部资产。

分部健康度矩阵:

高利润率低利润率
高增长明星分部需投入
低增长现金牛待转型/退出

经验阈值:增速 >20% 高增长投入期、10-20% 稳健、0-5% 成熟、负值警戒;利润率 >30% 高端、20-30% 强、10-20% 中等、<5% 单薄需规模。

重组红旗(警惕):

  1. 频繁变更口径——可能掩盖真实表现。
  2. 合并衰退分部——藏弱点。
  3. 变更时点恰逢经营问题——蹊跷。
  4. 披露减少——透明度下降。
  5. 更换上报 KPI——口径不可比。

对标的可比性调整: 分部定义、成本分摊方法、转移定价、合并抵销口径各家不同,做跨公司对比前需对齐。

分析技巧: 细读分部脚注(政策与变更都在此);自己复算利润率核对官方数字;关注营收构成迁移(反映战略优先级);季度数据注意季节性;与管理层分部指引对照。

互见

  • Octagon MCP 配置:参见源仓库 references/mcp-setup.md
  • 结果解读细则:参见源仓库 references/interpreting-results.md(分部构成、盈利能力评估、重组追踪、地区分部)。
  • 同领域可关联:SEC 财报/盈利分析类技能、估值(sum-of-parts)类技能。

采编自 OctagonAI/skills(MIT 许可),适配重写为中文技能大典条目。

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Keep looking

Skills are one crate of 326,546. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.