Skip to content

getting_started.md

Latest commit

 

History

History
142 lines (108 loc) · 3.19 KB

getting_started.md

File metadata and controls

142 lines (108 loc) · 3.19 KB

Getting Started with Agentify

Installation

pip install agentify-core

For development:

git clone https://github.com/fa8i/Agentify.git
cd Agentify
pip install -e .

Prerequisites

  • Python 3.10+
  • API Key from your chosen provider (OpenAI, DeepSeek, Gemini, etc.)

Environment Setup

Create a .env file in your project root:

OPENAI_API_KEY=your-key-here
# Or for other providers:
# DEEPSEEK_API_KEY=your-key-here
# GEMINI_API_KEY=your-key-here
# ANTHROPIC_API_KEY=your-key-here
# LOCAL_API_BASE=http://localhost:1234/v1
# LOCAL_API_KEY=api_key (dummy for local servers)

Your First Agent

import os
from dotenv import load_dotenv
from agentify import BaseAgent, AgentConfig, MemoryService, MemoryAddress
from agentify.memory.stores import InMemoryStore

load_dotenv()

# 1. Setup Memory
memory = MemoryService(store=InMemoryStore())
addr = MemoryAddress(conversation_id="my_first_chat")

# 2. Create Agent
agent = BaseAgent(
    config=AgentConfig(
        name="MyFirstAgent",
        system_prompt="You are a helpful assistant.",
        provider="provider",
        model_name="model_name",
        temperature=0.7,
    ),
    memory=memory,
    memory_address=addr
)

# 3. Chat (sync)
response = agent.run("Hello! Who are you?")
print(response)

# Async alternative:
# response = await agent.arun("Hello! Who are you?")

Streaming Responses

Enable streaming for real-time output:

agent = BaseAgent(
    config=AgentConfig(
        name="StreamAgent",
        system_prompt="You are a helpful assistant.",
        provider="provider",
        model_name="model_name",
        stream=True,  # Enable streaming
    ),
    memory=memory,
    memory_address=addr
)

# Get a sync generator
response = agent.run("Tell me a story")

# Stream the response (sync)
for chunk in response:
    print(chunk, end="", flush=True)

# Async streaming alternative:
# response = await agent.arun("Tell me a story")
# async for chunk in response:
#     print(chunk, end="", flush=True)

Adding Tools

Tools give your agent capabilities:

from agentify.extensions.tools import TimeTool, CalculatorTool

agent = BaseAgent(
    config=AgentConfig(...),
    memory=memory,
    memory_address=addr,
    tools=[TimeTool(), CalculatorTool()]  # Add tools here
)

response = agent.run("What time is it? Also calculate 15 * 23")
print(response)

Async Execution (Parallelism)

arun() enables non-blocking execution and parallel tool calls:

  1. Non-blocking execution: Your server stays responsive while waiting for the LLM.
  2. Parallel Tool Calls: If the agent needs multiple tools (e.g., getting weather for 3 cities), it executes them simultaneously, saving time.
import asyncio

async def main():
    # ... setup agent as above ...
    
    response = await agent.arun("Get weather for Tokyo, London, and NY")
    print(response)

# Run the async loop
asyncio.run(main())

Next Steps