StudyBuddy
101mlevydev/studybuddy
Converte provas e notas em guias de estudo com IA, com mapas conceituais e estratégia de aprendizado
Ver no GitHubCLI local em Python que transforma provas de universidade e anotações de aulas em guias de estudo estruturados e potencializados por IA. Gera dossiês por questão, mapas conceituais, guias de estratégia e questões práticas usando Claude, com rastreamento de custo em cada chamada.
Instalação
npx convez-ai 101mlevydev/studybuddy1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Preparação para exames universitários com análise automatizada de questões passadas
- Geração de material didático estruturado a partir de PDFs e apresentações
- Mapeamento de conceitos-chave e suas conexões para entendimento aprofundado
Veredito

Ferramenta bem arquitetada para automatizar criação de guias de estudo com pipeline robusto, custo tracking transparente e suporte a múltiplos formatos. Ideal para estudantes que querem otimizar tempo de preparação mantendo controle total dos gastos com LLM.
README
A tradução para o português deste README ainda está na fila.
StudyBuddy
StudyBuddy is a local Python CLI that turns your raw university exam papers and lecture notes into a structured, AI-powered study guide. It reads PDFs and slide decks, understands each exam question, and produces per-question dossiers, concept maps, a strategy guide that ranks the highest-leverage topics, and practice questions — using Anthropic's Claude for OCR, analysis, and synthesis. It handles Hebrew and English content.
The whole thing runs as an async pipeline you can stop and resume, with cost tracking on every LLM call so you always know what a run will spend.
Features
- Per-question dossiers — for every exam question: the concepts it tests, the solution approach, common traps, and the formulas involved.
- Concept maps — extracts the core concepts across a course and how they connect.
- Strategy guide — ranks topics by how high-leverage they are for the exam, so you know what to study first.
- Concept lectures — generates focused explanations (intuition, derivation, worked problems, common wrong models) per concept.
- Practice questions — generated from the patterns found in past papers.
- Resumable runs — a checkpoint store means a long run can be interrupted and continued without redoing work.
- Cost tracking — per-stage token usage and cost breakdown, with an upfront estimate command.
How it works
A staged pipeline: ingest -> segment -> categorize -> analyze notes -> per-question analysis -> synthesize -> render. Documents are parsed (pdfplumber / pypdf / python-pptx), state is indexed in SQLite, every step is validated with Pydantic models, and output is rendered to a self-contained HTML study guide.
Prerequisites
- Python 3.11+
- uv (package manager)
- An Anthropic API key — create one at https://console.anthropic.com
Installation
# Install uv if you don't have it
curl -LsSf https://astral.sh/uv/install.sh | sh
# Install all dependencies (creates .venv automatically)
uv sync --dev
# Configure your API key
cp .env.example .env
# edit .env and set ANTHROPIC_API_KEY=sk-ant-...
Usage
Put a course's exam papers and notes under courses/<course-name>/, then:
study-buddy --help
study-buddy estimate --course linear-algebra # preview the cost before running
study-buddy run --course linear-algebra # run the full pipeline
study-buddy run --course linear-algebra --stage 4
study-buddy run --course linear-algebra --resume
study-buddy status --course linear-algebra
Your courses/ directory holds personal exam papers and notes and is gitignored, so it is never committed.
Development
uv run pre-commit install # install git hooks (run once after cloning)
uv run pytest tests/ -v
uv run ruff check src/ tests/
uv run mypy src/
Hooks run automatically on git commit (ruff lint/format, mypy strict, unit tests). To run manually: uv run pre-commit run --all-files.
Documentation
Design and architecture docs live in docs/ (overview, pipeline, data model, prompts, outputs spec, tech stack, quality bar), and the full product design is in STUDYBUDDY_DESIGN.md.