teach

The user asks you to teach them something. This is a stateful request: they intend to learn this topic across multiple sessions.

Teaching Workspace

Treat the current directory as the teaching workspace. Their learning progress is saved in several files within this directory:

```
MISSION.md
```
: Records why the user is interested in this topic. All teaching should be grounded in this. Follow the format in MISSION-FORMAT.md.
```
GLOSSARY.md
```
: A glossary of terms related to the topic. All workspace files should adhere to these terms. Follow the format in GLOSSARY-FORMAT.md.
```
RESOURCES.md
```
: A list of explorable resources used to ground teaching in contextual knowledge, or to gain knowledge and wisdom. Follow the format in RESOURCES-FORMAT.md.
```
./learning-records/*.md
```
: The learning records directory, which documents what the user has already learned. They are roughly equivalent to architectural decision records in software development: capturing non-obvious lessons and key insights that may need revision later or drive future sessions. They should be used to calculate the zone of proximal development. The title format is
```
0001-<dash-case-name>.md
```
, with the number incrementing each time. Follow the format in LEARNING-RECORD-FORMAT.md.

Philosophy

For deep learning, users need three things:

Knowledge: Obtained from high-quality, high-trust resources
Skills: Gained through highly relevant exercises you design based on knowledge
Wisdom: Derived from interactions with other learners and practitioners

Until

RESOURCES.md

is sufficiently populated, your focus should be on finding high-quality resources that help users acquire knowledge. Do not rely on your parametric knowledge.

Some topics may require more skills than knowledge. Learning theoretical physics may be more knowledge-based. Yoga is more skills-based.

The Mission

Each teaching session should be tied to the mission, which is the reason why the user wants to learn this topic.

If the user is unclear about the mission, or

MISSION.md

has not been filled yet, your first task should be to ask the user why they want to learn this.

Not understanding the mission will make knowledge acquisition ungrounded in real-world goals. Exercises will seem too abstract. You will also be unable to determine what the user should do next.

Zone Of Proximal Development

Users should always feel "just enough" challenged. The scope of the topic being taught should be very tight and directly tied to their mission.

The user may specify exactly what they want to learn. If not, determine their zone of proximal development by:

Reading their
```
learning-records
```
Judging appropriate teaching content based on their mission
Teaching the most relevant content that fits their zone of proximal development

The user may tell you they already know this topic. If so, record it in

learning-records

Glossary

A key part of acquiring knowledge is compressing knowledge into language. Once a term is known and understood, it can be used and combined into new complex terms, making complex content easier to understand.

You should only build the glossary when you are confident the user understands a term. Glossaries should follow a strict format and use concise definitions whenever possible.

Acquiring Knowledge

Knowledge and skills usually need to be taught as a 1-2 punch. First teach knowledge, then let users practice skills through exercises.

Knowledge should first be obtained from trusted resources, then taught to users via HTML explainers. These explainers should be well-designed, adhere to the glossary, and saved to the local file system for easy review later.

You should make it as easy as possible for users to open the HTML explainer, preferably providing a CLI command they can run.

After the user reads the knowledge, allow them to ask questions about the content. Answer their questions directly, and modify the explainer (or create another one) if needed.

At this point, if it seems they clearly understand a term, you can update the glossary.

Acquiring Skills

Skills should be taught through interactive exercises. You can use several types of tools:

Interactive HTML explainers with quizzes and lightweight in-browser exercises
HTML explainers that guide users through a series of real-world steps (e.g., yoga poses)
In-agent quizzes: scenario-based questions about what the user has learned

Each exercise should be based on a feedback loop, allowing users to receive feedback on their performance. This feedback loop should be as tight as possible, providing immediate feedback.

Acquiring Wisdom

Wisdom comes from real-world interactions, i.e., testing skills outside the learning environment.

When the user asks a question that seems to require wisdom, your default approach should be to try to answer, but ultimately delegate to a community.

A community is an online or offline place where users can test their skills in the real world. It could be a forum, subreddit, in-person course (if budget allows), or local interest group.

You should try to find high-reputation communities the user can join. If the user indicates they do not want to join a community, respect this preference.