Tutorials

10xGraph tutorials are guided walkthroughs of the example code in the repository’s examples/ folder. They sit between the quickstart pages and the reference: each one explains a working script, covering agents, custom state, tools, streaming, MCP, memory, multimodal input, multi-agent handoff, skills, testing and evaluation. They are for developers who learn best by reading and running real code.

Start here

Begin with the Agent Class pattern, the smallest useful graph. Continue with Custom state and the Tool decorator to see how state and tools are modeled, then assemble the loop in the ReAct agent tutorial.

For real-time output, React streaming and Stop stream cover token streaming and cancellation. For connecting outside systems, see MCP ReAct agent and Memory. The Testing and Evaluation tutorials show how to check an agent before release, and Graceful shutdown covers stopping cleanly.

Tutorials teach the library. To run an agent as a service, with auth and deployment, use the how-to guides and the beginner path.

Tutorial track

flowchart LR
    A[Agent Class] --> B[Custom State]
    A --> C[Tool Decorator]
    A --> D[ReAct Agent]
    A --> E[Google GenAI Adapter]
    D --> F[ReAct Agent with Validation]
    D --> G[React Streaming]
    G --> H[Synchronous Streaming]
    H --> I[Stop Stream]
    I --> J[Dependency Injection]
    J --> K[MCP Server]
    K --> L[MCP Client]
    L --> M[MCP ReAct Agent]
    M --> N[GitHub MCP]
    N --> O[MCP File Download]
    O --> P[Memory]
    P --> Q[Qdrant Memory]
    Q --> R[Multimodal]
    R --> S[Multiagent]
    S --> T[Handoff]
    T --> U[Skills]
    U --> V[Skills Chat]
    V --> W[Testing]
    W --> X[Evaluation]
    X --> Y[Graceful Shutdown]

From examples

These tutorials are based on code in examples/:

• Agent Class Pattern
• Custom State
• Google GenAI Adapter
• Tool Decorator
• ReAct Agent
• ReAct Agent with Validation
• React Streaming
• Stop Stream

Advanced integrations

• Dependency Injection
• MCP Server
• MCP Client
• MCP ReAct Agent
• GitHub MCP
• Memory
• Multimodal
• Multiagent
• Handoff
• Skills
• Testing
• Evaluation
• Graceful Shutdown

How to use this section

If you are new to 10xGraph, follow the pages in the order shown above. The tutorials are designed to build on each other:

• Start with Agent Class Pattern to learn the smallest useful graph.
• Move to Custom State and Tool Decorator to understand how state and tools are modeled.
• Use ReAct Agent to assemble the core graph loop.
• Add ReAct Agent with Validation for safer input handling.
• Finish with the streaming tutorials to learn real-time output and cancellation patterns.
• Continue into the advanced section for MCP, memory, multimodal input, and multi-agent coordination.
• Finish with the sprint 9 tutorials to learn skills, testing, evaluation, and clean shutdown behavior.

Before you start

Most tutorials assume:

• Python 3.12 or later
• 10xGraph installed (pip install 10xgraph)
• Environment variables loaded from .env
• A provider key such as GEMINI_API_KEY for Google-based examples

Related docs

• First Python Agent
• State Graph
• Agents and Tools
• Streaming