Designing Skill-Driven Financial Analysis Agents with Claude, Python, MCP Connectors, and Automated Deliverables

This tutorial demonstrates how to build a skill-driven financial analysis agent using Anthropic's financial-services repository. It begins by cloning the repository, installing dependencies (anthropic, pandas, openpyxl, pyyaml, matplotlib), and mapping the repository structure including agent plugins, vertical plugins, partner integrations, and managed-agent cookbooks. The system parses SKILL.md files into a SkillRegistry, then constructs a SkillAgent class that injects selected financial playbooks into the Claude model via the Anthropic Messages API with tool-use loop. The agent executes a synthetic discounted cash flow valuation for a company called Meridian Software, computing enterprise value, equity value, and implied share price. It generates a WACC vs terminal growth sensitivity heatmap, performs comparable-company analysis with Excel output, drafts a private-equity investment committee memo, and inspects a managed-agent deployment specification. All numerical work is done via a Python execution tool, and final deliverables are saved to an 'outputs' directory.