用Claude和Python构建技能驱动的金融分析智能体

在本教程中，我们围绕Anthropic的金融服务库：https://github.com/anthropics/financial-services 构建一个高级工作流，并在纯Python中重现其基于技能的架构。我们首先安装所需的库，克隆该库，并以编程方式映射其代理、垂直插件、合作伙伴集成、托管代理手册和金融分析技能。然后，我们将库中的SKILL.md文件解析为可搜索的注册表，并构建一个可重用的SkillAgent，将选定的金融操作手册注入Anthropic Messages API，同时支持Python计算和文件生成的迭代工具使用循环。使用此架构，我们执行合成折现现金流估值，生成加权平均资本成本（WACC）和终值增长敏感度热图，进行可比公司分析并输出格式化Excel，起草私募股权投资委员会备忘录，并检查托管代理部署规范，而无需发送实时部署请求。

我们安装所需的Python库，克隆Anthropic的金融服务库，并准备Google Colab运行时以执行操作。我们从Colab Secrets、环境变量或安全的交互式提示中获取Anthropic API密钥。接着，我们初始化官方Anthropic SDK，并选择驱动金融分析工作流的Claude模型。

我们检查库的结构，识别其代理插件、垂直插件、合作伙伴集成、托管代理手册以及可用命令。我们定位MCP配置文件，并显示库中定义的外部金融数据连接器。然后，我们解析每个SKILL.md文件，提取其YAML元数据和方法论，并注册每个唯一技能以便可搜索访问。

我们定义工具，使Claude能够执行Python计算并在Colab环境中保存生成的交付成果。我们构建一个持久的Python命名空间，以便数值模型、表格和中间变量在多次代理操作中保持可用。然后，我们创建SkillAgent类，将选定的金融操作手册注入其系统提示，并管理Anthropic Messages API的工具使用循环。

我们提供合成的运营假设，并指示DCF技能代理构建五年未杠杆现金流估值。我们使用Python执行工具计算企业价值、股权价值、隐含股价以及二维敏感性矩阵。然后，我们提取结构化的敏感性结果，并通过热力图可视化WACC、终端增长率和隐含估值之间的关系。

我们提供合成的可比公司集合，并计算企业价值、企业价值与收入比（EV/Revenue）、企业价值与息税折旧摊销前利润比（EV/EBITDA）以及市盈率（P/E）倍数。我们将代理的结构化JSON响应转换为详细的可比公司及统计汇总DataFrame。然后，我们将分析结果导出到多工作表Excel工作簿，并应用专业的表头格式及自动列宽调整。

我们将投资委员会备忘录技能应用于一个假设的软件收购案例，并用Python计算支持的回报指标。我们将生成的备忘录保存为Markdown交付文件，并验证生成的文件是否存在于输出目录中。随后，我们检查托管代理手册，显示部署规格，并最终审阅生成的工件及可能的生产扩展方案。

通过完成本教程，我们实现了基于Colab的Anthropic金融服务技能及代理框架的实用近似，同时保持仓库在金融分析方法上的指导性方法。我们在单一可复用的工作流中结合了结构化技能发现、动态系统提示构建、持久Python执行、基于API的工具编排及自动化交付生成。我们还展示了同一代理架构如何支持多种金融用例，包括DCF估值、可比交易分析、敏感性测试、Excel报告以及投资委员会备忘录编制。从这里开始，我们可以通过加载更多技能、组合多个估值操作手册、通过MCP集成连接授权金融数据提供商以及将教程沙箱替换为Claude Code、Cowork或托管代理的受控生产运行时来扩展系统。

想与我们合作推广您的GitHub仓库、Hugging Face页面、产品发布或网络研讨会等？请联系我们：https://forms.gle/wbash1wF6efRj8G58

Sana Hassan 是 Marktechpost 的咨询实习生，也是 IIT 马德拉斯的双学位学生，他热衷于将技术和人工智能应用于解决现实世界的挑战。由于对解决实际问题充满兴趣，他为人工智能与现实生活解决方案的结合带来了新的视角。

构建一个自主事件场地运营商 [完整代码]：https://pxllnk.co/twdn5

谢谢！我们的团队会尽快与您联系 🙌

In this tutorial, we build an advanced workflow around Anthropic’s financial-services ：https://github.com/anthropics/financial-services repository and reproduce its skill-driven architecture in pure Python. We begin by installing the required libraries, cloning the repository, and programmatically mapping its agents, vertical plugins, partner integrations, managed-agent cookbooks, and financial analysis skills. We then parse the repository’s SKILL.md files into a searchable registry and construct a reusable SkillAgent that injects selected financial playbooks into the Anthropic Messages API while supporting an iterative tool-use loop for Python calculations and file generation. Using this architecture, we execute a synthetic discounted cash flow valuation, generate a WACC and terminal-growth sensitivity heatmap, perform comparable-company analysis with formatted Excel output, draft a private-equity investment committee memo, and inspect a managed-agent deployment specification without sending a live deployment request.

We install the required Python libraries, clone Anthropic’s financial-services repository, and prepare the Google Colab runtime for execution. We retrieve the Anthropic API key from Colab secrets, environment variables, or a secure interactive prompt. We then initialize the official Anthropic SDK and select the Claude model that powers the financial-analysis workflows.

We inspect the repository structure and identify its agent plugins, vertical plugins, partner integrations, managed-agent cookbooks, and available commands. We locate MCP configuration files and display the external financial data connectors defined in the repository. We then parse each SKILL.md file, extract its YAML metadata and methodology, and register every unique skill for searchable access.

We define tools that allow Claude to execute Python calculations and save generated deliverables inside the Colab environment. We build a persistent Python namespace so numerical models, tables, and intermediate variables remain available across multiple agent turns. We then create the SkillAgent class, inject selected financial playbooks into its system prompt, and manage the Anthropic Messages API tool-use loop.

We provide synthetic operating assumptions and instruct the DCF skill agent to construct a five-year unlevered cash-flow valuation. We use the Python execution tool to calculate enterprise value, equity value, implied share price, and a two-dimensional sensitivity matrix. We then extract the structured sensitivity results and visualize the relationship between WACC, terminal growth, and implied valuation through a heatmap.

We supply a synthetic peer set and calculate enterprise value, EV-to-revenue, EV-to-EBITDA, and price-to-earnings multiples. We convert the agent’s structured JSON response into detailed comparable-company and summary-statistics DataFrames. We then export the analysis to a multi-sheet Excel workbook and apply professional header formatting and automatic column sizing.

We apply the investment-committee memo skill to a hypothetical software buyout and calculate supporting return metrics with Python. We save the resulting memo as a Markdown deliverable and verify that the generated file exists in the output directory. We then inspect a managed-agent cookbook, display the deployment specification, and conclude by reviewing the generated artifacts and possible production extensions.

By completing this tutorial, we implement a practical Colab-based approximation of Anthropic’s financial-services skill and agent framework while preserving the repository’s methodology-driven approach to financial analysis. We combine structured skill discovery, dynamic system-prompt construction, persistent Python execution, API-based tool orchestration, and automated deliverable generation in a single reusable workflow. We also demonstrate how the same agent architecture supports multiple finance use cases, including DCF valuation, trading-comps analysis, sensitivity testing, Excel reporting, and investment committee memo preparation. From here, we extend the system by loading additional skills, combining multiple valuation playbooks, connecting to licensed financial data providers via MCP integrations, and replacing the tutorial sandbox with a governed production runtime in Claude Code, Cowork, or Managed Agents.

Need to partner with us for promoting your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar etc.? Connect with us ：https://forms.gle/wbash1wF6efRj8G58

Sana Hassan, a consulting intern at Marktechpost and dual-degree student at IIT Madras, is passionate about applying technology and AI to address real-world challenges. With a keen interest in solving practical problems, he brings a fresh perspective to the intersection of AI and real-life solutions.

Build an Agentic Event Venue Operator [Full Codes]：https://pxllnk.co/twdn5

Thanks! Our team will contact you soon 🙌