Ads Acquisition Agent

This project is a LangGraph-based template designed to audit Meta ads and webpage content to provide insights on ad performance and compliance.

Features

• Meta Ads auditing
• Webpage content analysis
• Data analysis of collected information
• Automated report generation
• Multi-agent collaboration with specialized experts

Project Structure

The application is built using LangGraph, a framework for building stateful, multi-agent workflows. It includes:

• Multiple specialized agents (Workshop Supervisor, Ads Expert, Web Expert, Data Analyzer)
• Tool nodes for data collection and processing
• A configurable workflow graph that coordinates the agents

Requirements

• Python 3.11+
• LangGraph
• LangChain
• OpenAI API key

Installation

1. Clone the repository:

git clone https://github.com/AIHackersLabs/ads-adquisition-agent-template
cd ads-adquisition-agent-template

2. Create a virtual environment:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

4. Create a .env file with your API keys:

OPENAI_API_KEY=your-openai-api-key
LANGSMITH_API_KEY=your-langsmith-api-key  # Optional, for tracing
LANGCHAIN_TRACING_V2=true  # Optional, for tracing
LANGSMITH_PROJECT=your-project-name  # Optional, for tracing

Running the Application

To run the application locally with LangGraph dev server:

langgraph dev

This will start the LangGraph server.

Using the API

You can interact with the application using the LangGraph SDK:

from langgraph_sdk import get_client

client = get_client(url="http://localhost:2024")

async for chunk in client.runs.stream(
    None,  # Threadless run
    "agent",  # Name of assistant. Defined in langgraph.json
    input={
        "messages": [{
            "role": "human",
            "content": "Audit this Meta ad: <URL>",
        }],
    },
    stream_mode="updates",
):
    print(f"Receiving new event of type: {chunk.event}...")
    print(chunk.data)
    print("\n\n")

Customization

To customize the workflow:

1. Modify graph.py to add new agents or tools
2. Update the graph structure by modifying the edges between nodes
3. Implement your own custom tools by updating the tool node functions

License

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