sec-edgar-agentkit

The SEC EDGAR Agent Kit provides a multi-framework toolkit for building AI agents that access and analyze SEC EDGAR data. It exposes LangChain integrations, an MCP-use interface for simple natural language queries, a Gradio-based interactive interface, and Dify integration options. Use this MCP-enabled server to expose actions such as company information lookups, filing searches and content extraction, financial data parsing, and insider trading analysis, all wired through the MCP protocol so you can connect frameworks like LangChain or MCP-use clients. To get started, connect your agent or tool to the MCP server URL provided by your deployment (often sec-edgar-mcp or a local endpoint) and enable the desired actions (companies, filings, financial, insiderTrading) as described in the configuration examples. Then call the available tools (e.g., sec_edgar_filing_search, sec_edgar_company_info, sec_edgar_xbrl_parse) through your chosen framework to perform searches, fetch filings, or analyze financial data.

Prerequisites:\n- Bun (for TypeScript/JavaScript development) and/or Node.js, Python 3.8+ for Gradio interfaces.\n- Access to a running sec-edgar-mcp MCP server.\n\nInstallation steps:\n1) Clone the repository:\n git clone https://github.com/stefanoamorelli/sec-edgar-agentkit\n cd sec-edgar-agentkit\n\n2) Install dependencies (monorepo):\n bun install\n\n3) Build all packages (if needed):\n bun run build\n\n4) Install individual packages (examples):\n # LangChain toolkit\n npm install @sec-edgar-agentkit/langchain\n\n # MCP-use natural language interface\n npm install @sec-edgar-agentkit/mcp-use\n\n # smolagents Python package (via pip)\n pip install sec-edgar-agentkit-smolagents\n\n5) Run or integrate with the MCP server: ensure sec-edgar-mcp is installed and running, then start the agentkit tooling or libraries that connect to your MCP server endpoint as shown in the Quick Start examples.

Notes and tips:\n- This repository is a monorepo of multiple packages optimized for different AI agent frameworks. There isn’t a single standalone MCP server entrypoint in this repo; instead, it provides libraries and integrations that connect to an MCP server such as sec-edgar-mcp. Make sure the MCP server is installed and reachable at runtime.\n- For Gradio and Python interfaces, follow the integration docs inside integrations/gradio and integrations/dify.\n- Environment variables and configuration are typically managed in your MCP setup and the specific integration you use (e.g., LangChain toolkit settings).\n- If you update packages, re-run bun install and bun run build to ensure consistency across the monorepo.