MMCPMarkets
PD

pingcy/deepagents-demo

API development

This page was translated automatically. Technical details may differ from the original. View the original

73 stars Quality 48 Trend 14
Repository Install

Agent Skills capability demonstration project based on LangGraph and DeepAgents frameworks.

Overview

Agent Skills capability demonstration project based on LangGraph and DeepAgents frameworks. This project demonstrates how to use the framework to build an agent system with capabilities. Through the Skills mechanism, agents can dynamically load and use various predefined skill modules, achieving more powerful and flexible task execution capabilities. - 🛠️ : Implements Anthropic's Agent Skills pattern, supporting dynamic loading and progressive disclosure of skills - 🔧 : Expands agent capabilities through SkillsMiddleware and ShellMiddleware - 🤖 : Supports collaboration between main agents and specialized sub-agents (such as search agents) - 📁 Supports virtual file systems for easy management of generated content - 🌐 Integrates Tavily and SearXNG search engines 1. : The agent only knows the name and description of the skill and only reads the full command when needed. 2. : - User-level Skills: ~/.deepagents/{AGENT_NAME}/skills/ - Project-level Skills: {PROJECT_ROOT}/agent/skills/ - : Structured web research skills, supporting multi-task parallel research - : Full-stack application template generator (FastAPI + React + Vite) - search - Universal web search (Tavily + SearXNG) - fetch_url - Retrieves URL content and converts it into Markdown - get_stock_price - Get historical A-share market data - get_technical_indicators - Calculate technical indicators - get_financial_statements - Get financial statements - get_stock_detailed_info - Get company details - get_stock_news - Get individual stock news - get_stock_research_report - Get institutional research reports - Python >= 3.13 - OpenAI API Key (or compatible API) - Tavily API Key (optional, for web search)

Overview (machine translated)

Agent Skills capability demonstration project based on LangGraph and DeepAgents frameworks.

Project Overview

This project demonstrates how to use the DeepAgents framework to build an agent system with Skills capabilities. Through the Skills mechanism, agents can dynamically load and use various predefined skill modules, achieving more powerful and flexible task execution capabilities.

Core Features

  • 🛠️ Skills System: Implements Anthropic’s Agent Skills model, supporting dynamic loading and progressive disclosure of skills
  • 🔧 Middleware Middleware Architecture: Extends agent capabilities through SkillsMiddleware and ShellMiddleware
  • 🤖 Sub-agent Collaboration: Supports collaboration between the main agent and specialized sub-agents (such as search agents).
  • 📁 File System Backend: Supports virtual file systems, making it easy to manage generated content
  • 🌐 Web Search Integration: Integrates Tavily and SearXNG search engines

Skills System Description

Skills is a mechanism that enables agents to possess professional capabilities:

  1. Progressive Disclosure: The agent only knows the name and description of the skill and only reads the full instructions when needed
  2. Table of Contents Structure:
  • User Level Skills: ~/.deepagents/{AGENT_NAME}/skills/
  • Project-level Skills: {PROJECT_ROOT}/agent/skills/

Built-in Skills examples

  • web-research: Structured web research skills supporting multi-task parallel research
  • fullstack-template-generator: Full-stack application template generator (FastAPI + React + Vite)

Project Structure

Toolkit

Search tools

  • search - General web search (Tavily + SearXNG)
  • fetch_url - Retrieves URL content and converts it to Markdown

Stock Data Tools (Optional)

  • get_stock_price - Get historical market data for A-shares
  • get_technical_indicators - Calculate technical indicators
  • get_financial_statements - Retrieves financial statements
  • get_stock_detailed_info - Get company details
  • get_stock_news - Get individual stock news
  • get_stock_research_report - Obtain institutional research reports

Data sources: AKShare, Baostock

Get Started Quickly

1. Environmental requirements

  • Python >= 3.13
  • OpenAI API Key (or compatible API)

This page was translated automatically. Technical details may differ from the original.

Recommended Tools

Try a different keyword or remove a filter.

Install

npx skillfish add pingcy/deepagents-demo
New to MCP? Start with the mcpskill.org guide →