EnkrateiaLucca

Solveit With Notebooks — Design skill for Claude Code

Design community

A teachable, SolveIt-style learning environment for Jupyter notebooks — Pólya's loop, dialogue not prompting, tiny steps, human-as-agent.

How to install Solveit With Notebooks

This entry records only its repository, not the path inside it, so there is no exact command to give. Open EnkrateiaLucca/solveit-with-notebooks and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Solveit With Notebooks does

A teachable, SolveIt-style learning environment for Jupyter notebooks — Pólya's loop, dialogue not prompting, tiny steps, human-as-agent. Inspired by Jeremy Howard's SolveIt.

Alternatives in Design

Brand Guidelines — Apply Anthropic's official brand colors and typography to artifacts 94.1k ★
Planning With Files — Claude Code skill implementing Manus-style persistent markdown planning — the workflow pattern behind the $2B 17.4k ★
Learning Output Style — Interactive learning mode for skill building 14k ★

README

🧩 SolveIt with Notebooks

A tiny, teachable re-creation of [Jeremy Howard's **SolveIt**](https://solve.it.com) that runs inside plain Jupyter notebooks — no special platform, just one importable file (`solveit.py`) and your normal notebook workflow.

Built for **learning and teaching** — research, Python basics, problem-solving — with AI as a thinking partner, not an answer machine.

![Python](https://img.shields.io/badge/python-3.12%2B-blue) ![License](https://img.shields.io/badge/license-MIT-green) ![Made with](https://img.shields.io/badge/made%20with-Jupyter-orange)

Why this exists

Most "AI in a notebook" setups encourage you to copy a big answer and move on. SolveIt does the opposite: it keeps **you** as the agent and turns the AI into a patient collaborator that works in small steps. This repo distills that idea into a handful of functions you can drop into *any* notebook.

It's built on **George Pólya's *How to Solve It* (1945)**, the same foundation SolveIt uses. The whole philosophy fits in four lines:

Principle	What it means	How this repo does it
Dialogue, not prompting	Talk with the AI over many small turns	`tutor` / `Dialogue` — a conversation that remembers
Pólya's loop	Understand → Plan → Do → Review	`solve("problem")` → `.understand() .plan() .step() .review()`
Tiny steps + feedback	1–2 lines, run, look, repeat	`.step()` proposes only the smallest next line
The human is the agent	AI proposes; you read, edit, run	`propose()` / `%%edit` never run code behind your back

Quickstart

# 1. an API key (the toolkit auto-detects which one you have)
export ANTHROPIC_API_KEY=sk-...     # uses Claude
# or
export OPENAI_API_KEY=sk-...        # uses GPT

# 2. dependencies
pip install anthropic openai ipywidgets jupyterlab
# optional extras:
#   pip install docling        # upload_pdf() — high-fidelity PDF → notebook (pymupdf is a lighter fallback)
#   pip