learn README

The project's README from github.com/amosblomqvist/learn. As one markdown file: README.md.

learn

video

My AI learning system from this video: How I Use AI to Learn Things.

This is a personal system I built for myself, shared as-is. Built as a pi configuration: the teaching philosophy encoded in a skill, a few small extensions, and agent definitions.

What's in it

Install

This repo is a .pi directory. From your learning project's root:

git clone https://github.com/amosblomqvist/learn .pi

Then open pi in that directory. (Or copy the pieces you want into your existing project config.)

Requirements

Notes

You can run the system without subagents. The main session does the teaching. You just lose the researcher (truth verification) and the generated visuals.

The teaching skill is written for one learner (me). Edit the skill to fit how you learn best.

learn (amosblomqvist/learn)

diagram
Searching Database…

3.7K ·GitHub
What is this?'learn' is a personal AI learning system built by the author for the 'pi' agent tool - it packages a teaching philosophy, a visualization skill, quiz/question extensions, and research subagents into a '.pi' configuration that turns pi into a tutor.
Is this a standalone app or a configuration I add to pi?It's not an app - it's a '.pi directory' containing skills, extensions, and agent definitions meant to be dropped into an existing pi-based project.
How do I add it to my learning project?Copy this repo into your project's root as the .pi directory (or copy just the pieces you want into your existing pi config), then open pi in that directory.
What do I need besides pi itself?You need a subagent implementation so pi can spawn the researcher and visual-maker agents - the author recommends 'pi-interactive-subagents' (tmux-only) since everything works out of the box with it, plus the bundled ask-user-question extension for popups.
Will it work with any subagent setup, or just the recommended one?Other implementations work too, but expect to adapt the agent definitions - e.g. agents/researcher.md lists 'safebash' in its tools, which is specific to pi-interactive-subagents, and popups only serialize correctly across the same ask-user-question implementation.
Who is this teaching style built for?It's a personal system written for the author's own way of learning - you're expected to edit the teach skill to match how you learn.
What do the quiz and md-log extensions add?Quiz turns a lesson into graded questions with instant checkmark/cross feedback, the correct answer, and an explanation; md-log links a markdown file to the session so you can keep running notes.
What happens if I skip the subagent setup entirely?The main session still teaches you directly, but you lose the researcher subagent's truth verification and the generated diagrams/SVGs.
What do the researcher, svg-maker, and mermaid-maker agents do?They are the subagents the system delegates to: researcher verifies facts, while svg-maker and mermaid-maker generate the diagrams used by the visualize skill.
Is the visualize skill always adding pictures?No, it only adds a diagram when an idea is genuinely clearer as a picture, aiming for a correct and minimal visual rather than decoration.
Size
0.3 MB
Stars
3.7K
Updated
CommitAug 25
Age
1mo
Commits · 90 Days