Skip to content

Course · Pi

The Pi course

This is a hands-on course, not a reference. Most coding agents hand you an opinionated machine and hide the wiring. Pi does the opposite: a bare core of four default tools, a deliberately compact system prompt, and no guardrails - everything else is yours to build. So this course is a build. You’ll spend a week with Pi turning an inherited side project - stash, a small self-hosted read-it-later service - into something you trust, and in the same motion you’ll assemble a Pi harness tailored to it: system rules, model routing, custom tools, safety gates, a verify loop, and a small agent team.

That double movement is the point. The model is the constant; the harness is the variable you control. Every module improves stash a little and your harness a lot, and the through-line question is the one worth carrying into any tool you use: how is the tool you have right now shaping what you believe is possible?

One enemy runs under all ten chapters: the context window is finite, and defaults spend it without asking. Every convenience another agent hands you - a permission prompt, a subagent it decided to spawn, a rule it loaded before you asked for it - is bought with tokens you didn’t choose to spend. Pi strips those defaults out, so every token in the window becomes one you put there on purpose. That trade is what this course actually teaches, one module at a time: watch a turn run with nothing hidden, read the window, spend it on rules instead of repeating yourself, route it to the model that’s worth the price, add a tool and own what it costs forever, rebuild only the guardrails worth their turns, load a skill on demand instead of always, split a job across windows and pay for the handoff, coordinate a small team and learn when that coordination isn’t worth it - and, in the last module, add the whole bill up and judge honestly whether it was worth paying.

If you want the spec-level “what is this primitive” view, that’s Foundations. This course is the other half: the muscle memory of building a harness with your own hands.

Ten chapters, in reading order. Each carries the stash project - and your growing harness - one stretch further.

Each chapter carries one running project - you’ll follow stash across a week of real work - and the lessons build on each other: each one picks up a problem the last one left you with, adds a piece to your harness, and hands off to the next.

So you get the most out of a chapter by reading it in order. But each lesson still stands on its own if you drop in from a search with one specific problem in front of you - you’ll just miss the running story.

What you won’t find is a fixed template stamped onto every lesson. Some need a long worked transcript, some a single command and a warning, some a digression into why Pi is built the way it is. The shape follows the content, the way a good book varies its chapters.