Tutorials¶

Step-by-step guides for building real agents with AgentFlow. Start with Beginner if you're new, or jump into any section once you have the basics.

Prerequisites¶

Before starting tutorials, make sure you've completed Getting Started:

AgentFlow installed (pip install 10xscale-agentflow)
An LLM provider library installed (google-genai or openai)
API key configured in .env

Beginner Path¶

Work through these in order for the smoothest learning curve:

Tutorial	What You Build	Key Skills
1. Your First Agent	A weather assistant with personality	System prompts, Agent class, basic workflow
2. Adding Tools	An agent that calls Python functions	ToolNode, conditional routing, tool execution
3. Chat with Memory	A persistent multi-turn chatbot	Checkpointers, thread IDs, conversation state

Total time: ~60 minutes

Building Agents¶

Deepen your Agent class knowledge and learn production patterns:

Tutorial	Focus
Agent Class Deep Dive	Configuration, tool filtering, context trimming, streaming
Tool Decorator & Filtering	Organize tools with metadata and tags
Multi-Agent Handoff	Delegate tasks between specialized agents
Input Validation	Sanitize inputs and protect against prompt injection

ReAct Pattern (Reasoning + Acting)¶

The ReAct pattern powers most production agents — reason, act with tools, observe results, repeat:

Tutorial	Focus
ReAct with Agent Class	Simplest setup — ReAct in under 30 lines
Custom ReAct (Advanced)	Build ReAct from scratch with custom async functions
Dependency Injection	Inject services and config into nodes with InjectQ
MCP Integration	Connect to Model Context Protocol servers
Streaming	Stream tokens in real-time to clients
Unit Testing	Test agents without real LLM API calls

Memory & Storage¶

Persist agent state and enable long-term memory:

Tutorial	Focus
Long-Term Memory	Persist memories across sessions using the Store API
Mem0 Store	Managed semantic memory with Mem0
Qdrant Vector Store	Vector database integration for similarity search
Embedding Store	Work with vector embeddings in agent workflows

Retrieval & Reasoning¶

Ground agents in external knowledge and complex multi-step reasoning:

Tutorial	Focus
RAG (Retrieval-Augmented Generation)	Ground agents in external documents with semantic search
Plan-Act-Reflect Pattern	Agents that plan, execute, and self-evaluate their work

Reference Docs¶

For API details on any class or method, see the Reference section:

Start here: Tutorial 1 — Your First Agent →