Skip to content

QUICKSTART.md

Latest commit

 

History

History
251 lines (178 loc) · 4.98 KB

QUICKSTART.md

File metadata and controls

251 lines (178 loc) · 4.98 KB

Quick Start Guide

Get up and running with the Distributed Multi-Agent Communication Platform in minutes.

Prerequisites

  • Python 3.9 or higher
  • Redis (optional, but recommended)
  • Gemini API key

Installation

1. Clone and Install

# Navigate to project directory
cd distributed_agent_platform

# Install dependencies
pip install -r requirements.txt

2. Configure Environment

# Copy example environment file
cp env.example .env

# Edit .env and add your Gemini API key
# GEMINI_API_KEY=your-key-here

3. Start Redis (Optional)

If you have Redis installed:

# macOS (Homebrew)
brew services start redis

# Linux
sudo systemctl start redis

# Windows
# Download and run Redis from https://redis.io/download

If Redis is not available, the system will fall back to in-memory queues.

4. Run the Platform

python main.py

The platform will start on http://localhost:8000

Using the Platform

Access the Dashboard

Open your browser to http://localhost:8000 to access the web dashboard.

Default Agents

The platform automatically creates four default agents on startup:

  • Researcher Agent (researcher_1): Specialized in information gathering
  • Analyst Agent (analyst_1): Performs data analysis
  • Coordinator Agent (coordinator_1): Orchestrates workflows
  • Knowledge Agent (knowledge_1): Manages knowledge base

Send a Message

Using the dashboard:

  1. Enter "researcher_1" in "From Agent ID"
  2. Enter "knowledge_1" in "To Agent ID"
  3. Select "Query" as message type
  4. Enter your query: "What is machine learning?"
  5. Click "Send Message"

Using the API

List Agents

curl http://localhost:8000/agents

Send a Message

curl -X POST http://localhost:8000/messages/send \
  -H "Content-Type: application/json" \
  -d '{
    "from_agent": "researcher_1",
    "to_agent": "analyst_1",
    "content": "Please analyze this data: Sales increased 20%",
    "message_type": "task"
  }'

Get Metrics

curl http://localhost:8000/metrics

WebSocket Connection

Connect an agent via WebSocket:

import asyncio
import websockets
import json

async def connect_agent():
    uri = "ws://localhost:8000/ws/researcher_1"
    async with websockets.connect(uri) as websocket:
        # Send a message
        message = {
            "id": "msg_1",
            "from_agent": "researcher_1",
            "to_agent": "analyst_1",
            "type": "text",
            "content": "Hello from WebSocket!"
        }
        await websocket.send(json.dumps(message))
        
        # Receive response
        response = await websocket.recv()
        print(json.loads(response))

asyncio.run(connect_agent())

Python API Usage

Create and Register an Agent

from backend.agents.researcher_agent import ResearcherAgent
from backend.core.agent_manager import AgentManager
from backend.services.vector_service import VectorService

# Initialize services
vector_service = VectorService()
await vector_service.initialize()

# Create agent
researcher = ResearcherAgent(
    agent_id="my_researcher",
    name="My Researcher",
    vector_service=vector_service
)

# Register with manager
agent_manager = AgentManager()
await agent_manager.register_agent(
    researcher.agent_id,
    researcher.name,
    researcher.agent_type,
    researcher.get_capabilities()
)

Send Messages Between Agents

from backend.models.message import Message, MessageType

# Create message
message = Message(
    id="msg_1",
    from_agent="researcher_1",
    to_agent="analyst_1",
    type=MessageType.TASK,
    content="Analyze this data: ..."
)

# Send via agent manager
response = await agent_manager.send_message(
    from_agent="researcher_1",
    to_agent="analyst_1",
    content="Analyze this data: ...",
    message_type=MessageType.TASK
)

Use Knowledge Base

# Add knowledge
doc_id = await vector_service.add_knowledge(
    content="Python is a programming language.",
    metadata={"category": "programming"}
)

# Search knowledge
results = await vector_service.search(
    query="What is Python?",
    n_results=5
)

Testing

Run the test suite:

pytest tests/

Run with coverage:

pytest --cov=backend --cov-report=html tests/

Troubleshooting

Redis Connection Error

If you see Redis connection errors, the system will automatically fall back to in-memory queues. For production, ensure Redis is running.

Gemini API Errors

Make sure your GEMINI_API_KEY is set in the .env file and is valid.

Port Already in Use

If port 8000 is in use, change it in .env:

SERVER_PORT=8001

Vector Database Errors

If ChromaDB initialization fails, check that the data/vectorstore directory is writable.

Next Steps