Quick Start Guide
Get up and running with AA Kit in under 5 minutes. This guide will walk you through creating your first agent.
Prerequisites
- Python 3.8 or higher
- An OpenAI or Anthropic API key
- Basic Python knowledge
1Install AA Kit
bash
pip install aa-kit2Create your first agent
python
from aakit import Agent
# Create an agent in 3 lines
agent = Agent(
name="assistant",
instruction="You are a helpful AI assistant",
model="gpt-4"
)
# Chat with your agent - no async/await needed!
response = agent.chat("What can you help me with?")
print(response)
# Or use async when you need it
response = await agent.achat("What can you help me with?")
print(response)3Add tools to your agent
python
from aakit import Agent
# Define a simple tool
def search_database(query: str) -> str:
"""Search our database for information."""
# Your search logic here
return f"Found results for: {query}"
# Create agent with tools
agent = Agent(
name="researcher",
instruction="You help users find information",
model="gpt-4",
tools=[search_database], # Automatically converted to MCP
reasoning="react" # Use ReAct pattern for tool usage
)
# Agent can now use tools - synchronous by default
response = agent.chat("Find information about Python")
# Or async version
response = await agent.achat("Find information about Python")4Compose agents together
python
from aakit import Agent
# Create specialized agents
researcher = Agent("researcher", "You research topics", "gpt-4")
writer = Agent("writer", "You write articles", "claude-3")
# Create a coordinator that uses other agents
coordinator = Agent(
name="coordinator",
instruction="You coordinate research and writing tasks",
model="gpt-4",
tools=[researcher, writer] # Agents as tools!
)
# Coordinate complex tasks - sync or async, your choice!
response = coordinator.chat(
"Research AI safety and write a short article about it"
)
# Or use async for concurrent operations
response = await coordinator.achat(
"Research AI safety and write a short article about it"
)5Have a conversation
python
from aakit import Agent
# Create an agent with memory
agent = Agent(
name="assistant",
instruction="You are a helpful AI assistant",
model="gpt-4",
memory="memory://" # Enable conversation memory
)
# Have a contextual conversation
with agent.conversation() as chat:
# Send messages - agent maintains context
chat.send("I'm learning Python")
chat.send("What should I learn first?") # Knows you're learning Python
chat.send("Can you show me an example?") # Knows the context
# Save the conversation
chat.save("learning_session.json")
# Or use interactive mode for a terminal chat
agent.interactive() # Opens REPL-like interface6Serve as MCP server
python
from aakit import Agent
# Create your agent
agent = Agent("assistant", "You are helpful", "gpt-4")
# Serve as MCP server - compatible with Claude Desktop!
agent.serve_mcp(
port=8080,
name="My Assistant",
description="A helpful AI assistant"
)
# Now accessible at http://localhost:8080
# Can be used by Claude Desktop or other MCP clients🎉 Congratulations!
You've just learned the core features of AA Kit:
- Creating agents with just 3 lines of code
- Adding tools that automatically become MCP-compatible
- Composing agents together for complex tasks
- Serving agents as MCP servers
Configuration
Set your API keys as environment variables:
# For OpenAI models
export OPENAI_API_KEY="your-api-key"
# For Anthropic models
export ANTHROPIC_API_KEY="your-api-key"
# Or use a .env file
echo "OPENAI_API_KEY=your-api-key" >> .envProduction Features
AA Kit comes with 10 production features built-in. Here's how to use them:
python
from aakit import Agent, ProductionConfig
# Configure production features
config = ProductionConfig(
rate_limit=100, # 100 requests per minute
cache_ttl=3600, # Cache for 1 hour
timeout=30, # 30 second timeout
retry_max=3, # Retry up to 3 times
)
agent = Agent(
name="production_agent",
instruction="You are a production-ready assistant",
model="gpt-4",
config=config
)