What is it?
What it is
A personal AI learning system built as a pi configuration, featuring teaching and visualization skills, interactive extensions (ask-user-question, quiz, md-log, visual-tools), and subagent definitions (researcher, svg-maker, mermaid-maker).
Why it exists
Created as a personal tool for the author's own learning, based on their YouTube video 'How I Use AI to Learn Things', then shared open-source so others can adapt the teaching philosophy and components to their own learning style.
Who should use it
Developers who want to set up an AI-assisted personal learning system using the `pi` coding agent, especially those comfortable customizing skill files and agent definitions to match their own learning style.
Who should avoid it
Users without the `pi` agent installed, those unwilling to install tmux and configure a subagent implementation, and learners looking for a ready-to-use, one-size-fits-all educational tool rather than a personal system meant to be edited and tailored.
How it works
A quick walkthrough in plain English
How learn works
Step 1 of 3
You interact with it
Open learn, send a request, or connect it to your stack.
Features
Advantages
- Strong community interest reflected by 810 GitHub stars
- Modular, plug-and-play design: components can be used together or individually
- Active development demonstrated by short, clear commit history and focused structure
- Clear separation of concerns between teaching, visualization, quizzing, and research
- Extensible to different subagent implementations, not locked to a single provider
- Includes bundled ask-user-question extension ensuring popup serialization works correctly
- Encourages personalization — skills are editable to fit individual learning styles
- Demonstrated via an associated explanatory video for easy onboarding
Disadvantages
- No license specified, which creates uncertainty about reuse and redistribution rights
- No listed topics/tags on the repository, reducing discoverability
- Requires the external pi runtime and a subagent implementation for full functionality
- Some tools referenced (e.g., safe_bash in agents/researcher.md) are extension-specific and require adaptation
- Documentation is minimal; deeper usage patterns and customization details are not provided
- Heavy reliance on tmux for the recommended subagent setup limits cross-platform flexibility
- Quality of generated diagrams and research depends entirely on the underlying LLM and subagent behavior
- Originally built for a single learner, so it may not generalize well without further customization
Installation
native
Clone the repository as a `.pi` directory from your learning project's root: ```bash 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: - [pi](https://github.com/earendil-works/pi) - A subagent implementation (recommended: [pi-interactive-subagents](https://github.com/amosblomqvist/pi-interactive-subagents), tmux only) - `ask-user-question` — use the copy bundled here
FAQ
What is the learn repository and how do I install it?
It is a personal AI learning system configured as a .pi directory. Install by running `git clone https://github.com/amosblomqvist/learn .pi` in your learning project's root directory, then open pi in that directory.
What are the requirements to run the learn system?
You need the pi framework from https://github.com/earendil-works/pi and a subagent implementation like pi-interactive-subagents for full functionality. The ask-user-question extension bundled here is also required and should replace any existing copy to ensure UI lock compatibility.
Can I run the system without subagents?
Yes, the main session handles teaching without subagents, but you lose the researcher (truth verification) and generated visuals. The teaching skill can be edited to fit your learning style.
What extensions are included and what do they do?
The system includes ask-user-question (UI popup questions), quiz (graded questions with instant feedback), md-log (link markdown files to sessions), and visual-tools (tools for visualization subagents).
What do the researcher, svg-maker, and mermaid-maker agents do?
These subagents are delegated by the system: researcher verifies truth via safe bash, svg-maker generates SVG diagrams, and mermaid-maker creates Mermaid diagrams. They enable research and visualization when subagents are active.
How customizable is the teaching philosophy?
The teaching skill is designed for one learner and is located in skills/teach/. You should edit it to fit how you learn best, as it is personalized for the author's preferences.
Featured in Videos
YouTube tutorials and walkthroughs for learn
Alternatives
Similar projects ranked by category, topics, and text overlap.
