Get started

This path takes you from a new project to a Python agent served by the AgentFlow API, tested in the hosted playground, and called from TypeScript.

It intentionally avoids advanced concepts. The goal is one working app first.

Golden path

Step	Page	Outcome
1	What is AgentFlow?	Understand what AgentFlow provides before installing it.
2	Installation	Install the Python library, CLI, and TypeScript client.
3	First Python Agent	Run a minimal graph locally.
4	Expose with API	Serve the graph with agentflow api.
5	Connect Client	Call the API with AgentFlowClient.
6	Open Playground	Start the API and hosted playground with agentflow play.

Prerequisites

You need:

• Python 3.12 or newer.
• Node.js 20 or newer if you want to run the TypeScript client example.
• A terminal where you can install Python and npm packages.

The first local graph example does not call an LLM, so you can verify the framework before adding provider keys.

How the pieces connect

The Python library owns the graph runtime. The CLI creates and serves a project around that graph. The TypeScript client and hosted playground both talk to the same local API server.

Commands you will use

pip install 10xscale-agentflow 10xscale-agentflow-cli
npm install @10xscale/agentflow-client
agentflow init
agentflow api
agentflow play

What you will have

By the end, you will have:

• A local graph/react.py file that exports app.
• An agentflow.json file that points the API to that graph.
• A running API server on http://127.0.0.1:8000.
• A hosted playground session opened by agentflow play.
• A minimal TypeScript AgentFlowClient call.

Next step

Start with What is AgentFlow?.