Sparsity Tutorials: Main Guide

Welcome! This repository is a collection of hands-on tutorials to help you learn how the Sparsity infrastructure works and how to build secure, tool-using AI agents on top of it.

---

⚙️ Requirements

Before you begin, make sure you have the following:

• OpenAI API key (or Gemini API key)
• Python >= 3.10
• Docker
• make
• (optional) venv (Python virtual environment)

---

For the python environment, we recommend you to create a python venv using the following command and activate it.

python -m venv .venv
source .venv/bin/activate

This will ensure the installed libraries/packages persist across the tutorial and take care of required libraries once you install them using pip install.

Below is the recommended order for working through the tutorials. Each step introduces new features and concepts, and each project has its own detailed Tutorial.md.

---

📚 Tutorial Index

1. tee-client

   • Learn: The basics of secure client-server communication using TEE (Trusted Execution Environment) and attestation.
   • Features: Simple encrypted chat, attestation verification, basic TEE API usage.
   • Start here: tee-client/Tutorial.md

2. coin-price-bot/local

   • Learn: How to build a local tool-using AI agent that can fetch and summarize real-world data (e.g., coin prices) using external APIs.
   • Features: Tool-using agent, API integration, prompt engineering, local development.
   • Go to: coin-price-bot/local/Tutorial.md

3. agent-mcp/local

   • Learn: How to build a more advanced, modular tool-using agent with multi-step reasoning and tool invocation.
   • Features: Modular agent design, multi-step tool use, advanced prompt workflows.
   • Go to: agent-mcp/local/Tutorial.md

4. coin-price-bot/sim-tee

   • Learn: How to deploy the coin-price-bot agent inside a simulated TEE, with secure proxying, DNS routing, and multi-step retrieval-augmented generation (RAG).
   • Features: TEE simulation, secure proxy, DNS interception, multi-step RAG agent, containerized deployment.
   • Go to: coin-price-bot/sim-tee/Tutorial.md

5. agent-mcp/sim-tee

   • Learn: How to deploy a modular, multi-tool agent inside a simulated TEE, with secure attestation, toolkits, and advanced agent orchestration.
   • Features: Secure TEE agent, attestation, toolkits, multi-step orchestration, containerized deployment.
   • Go to: agent-mcp/sim-tee/Tutorial.md

---

How to Use This Guide

• Work through the tutorials in order for the best learning experience.
• Each tutorial builds on the previous one, introducing new infrastructure, security, and agent features.
• For each project, follow the instructions in its Tutorial.md.

Happy hacking!