What is A2a With Mcp

a2a_with_mcp is a multi-agent system designed for market research and financial analysis, utilizing large language models (LLMs) and specialized agents to process user requests and gather information from various data sources through the Model Context Protocol (MCP).

Use cases

Use cases include conducting market research, performing financial analysis, analyzing sentiment on social media regarding cryptocurrencies, and comparing competitor performance.

How to use

To use a2a_with_mcp, users interact with the User Interface (UI) to submit requests. The Orchestrator Agent processes these requests, delegates tasks to specialized agents, and synthesizes the results into a final output.

Key features

Key features include an Orchestrator Agent for task management, specialized agents for financial analysis and sentiment analysis, connections to multiple data sources, and A2A-compliant FastAPI servers for agent communication.

Where to use

a2a_with_mcp can be used in financial institutions, market research firms, and any organization requiring in-depth analysis of market trends, financial data, and sentiment analysis related to specific assets.

A2A + MCP: Multi-Agent Financial Analysis System

This repository demonstrates a practical, real-world implementation of Google’s Agent-to-Agent (A2A) protocol combined with Anthropic’s Model Context Protocol (MCP). Together, these open standards enable seamless interoperability between modular AI agents.

🌟 What Does This Project Do?

We built a multi-agent system designed for financial market research, sentiment analysis, data scraping, and data visualization. Users interact through a streamlined UI, asking questions about crypto or market trends. Under the hood, requests are delegated to specialized agents via A2A, each powered by contextual data from MCP servers.

📌 Architecture Overview

• UI Orchestrator (A2A Client): Accepts user queries and forwards them using the A2A protocol.

• Remote Orchestrator (A2A Server): Delegates incoming tasks to specialized agents.

• Specialized Agents:

  • Financial Agent: Fetches crypto prices from local (Postgres MCP) and remote (CoinCap MCP) sources.
  • Sentiment Agent: Analyzes market sentiment using the Reddit MCP server.
  • Scraper Agent: Gathers additional market data through a Web Scraper MCP.
  • Visualization Agent: Prepares and generates visual charts.

All components communicate via standardized JSON-RPC (A2A) and access data via MCP.

🚦 Getting Started

Clone the repository:

git clone https://github.com/anshulLuhsna/a2a_with_mcp.git

cd a2a_with_mcp

Note: Each agent (and each MCP server) has its own requirements.txt. This keeps dependencies lightweight and fully modular.

Install dependencies per component—e.g.:

Financial Agent deps

pip install -r agents/financial/requirements.txt

Sentiment Agent deps

pip install -r agents/sentiment/requirements.txt

Scraper Agent deps

pip install -r agents/scraper/requirements.txt

Visualization Agent deps

pip install -r agents/visualization/requirements.txt

Remote Orchestrator deps

pip install -r agents/orchestrator/requirements.txt

UI dependencies

pip install -r demo/ui/requirements.txt

(Feel free to create a virtualenv or conda env first.)

🚀 Running the Demo

Run Agents:

Manually start each agent, e.g.

python -m agents.orchestrator
python -m agents.financial
python -m agents.sentiment
python -m agents.scraper
python -m agents.visualization

Start the UI:

cd demo/ui
uv run main.py

Open the UI at http://localhost:8501

📚 Resources & Documentation

• A2A Protocol (Google)
• MCP Documentation (Anthropic)

📜 License

This project is licensed under the MIT License. See LICENSE for details.