Beginner path

This path builds a working agent step by step. Each page teaches one concept, shows a complete code example, and tells you what to try next.

By the end you will have an agent that:

• Calls tools safely
• Persists conversation state with a checkpointer
• Runs behind an HTTP API
• Can be tested in the hosted playground
• Can be called from a TypeScript application

Prerequisites

Install AgentFlow before starting:

pip install 10xscale-agentflow
pip install 10xscale-agentflow-cli

Learning track

Step	Page	What you build
1	Mental model	Understand graph, state, message, and agent boundaries
2	Your first agent	Compile and run a single-node workflow
3	Add a tool	Give the agent a callable function
4	Add memory	Persist conversation state across calls
5	Run with the API	Expose the agent over HTTP
6	Test with the playground	Inspect requests with agentflow play
7	Call from TypeScript	Connect a frontend or Node.js client

How each page is structured

Every page in this path includes:

• A brief explanation of the concept
• A complete, runnable code example
• Expected output
• A "What you learned" section
• One clear next step

Start with the Mental model page.