Pi Agent Development · Course Overview
This course has 23 chapters, based on the official source documentation for @earendil-works/pi-agent-core and @earendil-works/pi-ai, building production-grade agents from scratch with TypeScript.
What You Will Learn
| Stage | Chapters | Content |
|---|---|---|
| Beginner | 01–05 | pi-ai unified LLM API, model routing, first Agent, event streaming |
| Intermediate | 06–10 | Tool definition & invocation, message transformation, context transformation, auth parsing |
| Advanced | 11–15 | Thinking/reasoning, images, error handling, custom providers, testing |
| Production | 16–20 | Cross-provider handoffs, session persistence, browser usage, comprehensive projects |
| Extensions | 21–23 | Telemetry observability, pi-tui terminal UI, three-chapter capstone project |
Official Reference Sources
| Source | Link |
|---|---|
| pi-agent-core package | packages/agent |
| pi-ai package | packages/ai |
| pi-telemetry package | packages/telemetry |
| pi-tui package | packages/tui |
| Pi website | https://pi.dev |
Prerequisites
- TypeScript/Node.js basics; it is recommended to complete the Pi Coding Agent Course on this site first;
- At least one LLM Provider API Key (Anthropic/OpenAI/DeepSeek, etc.).
👉 Start with Chapter 1.