acp-code

Summary Generator multi-agent workflow with ACP

A simple demonstration of the Agent Communication Protocol (ACP), showcasing how two agents built using different frameworks (CrewAI and Smolagents) can collaborate seamlessly to generate and verify a research summary.

---

Setup and Installation

1. Install Ollama:

   # Setting up Ollama on linux
   curl -fsSL https://ollama.com/install.sh | sh
   
   # Pull the Qwen2.5 model
   ollama pull qwen2.5:14b

2. Install project dependencies:

   Ensure you have Python 3.10 or later installed on your system.

   First, install uv and set up the environment:

   # MacOS/Linux
   curl -LsSf https://astral.sh/uv/install.sh | sh
   
   # Windows
   powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

   Install dependencies:

   # Create a new directory for our project
   uv init acp-project
   cd acp-project
   
   # Create virtual environment and activate it
   uv venv
   source .venv/bin/activate  # MacOS/Linux
   
   .venv\Scripts\activate     # Windows
   
   # Install dependencies
   uv add acp-sdk crewai smolagents duckduckgo-search ollama

You can also use any other LLM providers such as OpenAI or Anthropic. Create a .env file and add your API keys

OPENAI_API_KEY=your_openai_key
ANTHROPIC_API_KEY=your_anthropic_key

Usage

Start the two ACP servers in separate terminals:

# Terminal 1
uv run crew_acp_server.py

# Terminal 2
uv run smolagents_acp_server.py

Run the ACP client to trigger the agent workflow:

uv run acp_client.py

Output:

A general summary from the first agent

A fact-checked and updated version from the second agent

📬 Stay Updated with Our Newsletter!

Get a FREE Data Science eBook 📖 with 150+ essential lessons in Data Science when you subscribe to our newsletter! Stay in the loop with the latest tutorials, insights, and exclusive resources. Subscribe now!

---

Contribution

Contributions are welcome! Please fork the repository and submit a pull request with your improvements.