Getting Started¶
Welcome to Arshai! This section will get you up and running with the framework in minutes.
Quick Start Guides
- Installation
- Quickstart Guide
- Comprehensive Agent Guide
- Build Your First Custom Agent
- What We’ll Build
- Prerequisites
- Step 1: Set Up the Environment
- Step 2: Design Your Agent
- Step 3: Implement the Agent Core
- Step 4: Implement Complexity Analysis
- Step 5: Implement Processing Strategies
- Step 6: Create the Agent Instance
- Step 7: Test Your Agent
- Step 8: Make It Interactive
- Complete Example
- Key Concepts Demonstrated
- Next Steps
- Testing Your Agent
What is Arshai?¶
Arshai is a framework for building agentic AI systems. It provides:
Layer 1: Standardized LLM client interfaces
Layer 2: Agent building blocks with BaseAgent
Layer 3: Patterns for composing agents into systems
The framework follows these principles:
Direct Control: You explicitly create and configure components
Building Blocks: Framework provides foundations, you build solutions
Progressive Complexity: Start simple, scale to sophisticated systems
Choose Your Path¶
- 5 Minutes - Quickstart
Jump straight into code with Quickstart Guide. Create your first agent and start chatting.
- 30 Minutes - Comprehensive Guide
Read Comprehensive Agent Guide for a complete overview of agent capabilities and patterns.
- 1 Hour - Build Your First Agent
Follow Build Your First Custom Agent to build a production-ready agent step-by-step.
Quick Example¶
Here’s the simplest possible agent:
from arshai.agents.base import BaseAgent
from arshai.core.interfaces.iagent import IAgentInput
from arshai.core.interfaces.illm import ILLMInput
class MyAgent(BaseAgent):
async def process(self, input: IAgentInput) -> str:
llm_input = ILLMInput(
system_prompt=self.system_prompt,
user_message=input.message
)
result = await self.llm_client.chat(llm_input)
return result['llm_response']
Prerequisites¶
- Python Version
Python 3.8 or higher is required.
- API Keys
You’ll need an API key from one of:
OpenAI
Azure OpenAI
Google Gemini
OpenRouter
- Installation
Install Arshai via pip:
pip install arshai
What You’ll Learn¶
- Foundation Concepts
Creating agents by extending BaseAgent
Processing inputs and returning responses
Configuring LLM clients
- Core Patterns
Adding tools to agents
Managing conversation memory
Handling errors gracefully
- System Building
Composing multiple agents
Orchestration patterns
Building complete applications
Framework Philosophy¶
Arshai believes in:
Developer Authority: You control every aspect
Explicit Over Implicit: No hidden magic
Composition Over Monoliths: Build from simple pieces
Progressive Enhancement: Start simple, add complexity as needed
Next Steps¶
Install Arshai: Follow Installation
Try the Quickstart: Get hands-on in Quickstart Guide
Read the Guide: Understand concepts in Comprehensive Agent Guide
Build an Agent: Create something real in Build Your First Custom Agent
After getting started, explore:
Agents (Layer 2) - Deep dive into agents
Building Systems (Layer 3) - Compose agents into systems
LLM Clients (Layer 1) - Understand LLM integration
Welcome to the Arshai community!