Most study-note generators start with a textbook and summarize whatever looks important. The summarize-pdfs project starts with a different signal: exam papers. It uses questions from exams to guide retrieval from indexed OpenStax textbooks, then synthesizes study material around the topics students are likely to need.
The pipeline separates the work into commands: initialize the workspace, index textbook PDFs, parse exam papers, generate notes, and export the results. A combined produce command runs the full path. Outputs include structured JSON, a complete topic-organized guide, and a shorter set of quick notes with formulas and definitions.
Use exams as retrieval guidance
The key design choice is that exam papers act as ground truth for what deserves attention. Instead of producing a dump of question-by-question answers, the prompts aim for cohesive notes organized by topic. The project also has separate output modes: a fuller explanation and a compact revision sheet.
This is a study aid, not a replacement for the source textbook or an answer key. Its usefulness depends on the quality of the PDFs, parsing, retrieval, and generated synthesis. The repository keeps those steps inspectable and lets you run a small batch before processing a whole collection.
To try it, follow the summarize-pdfs quick start: initialize the project, add textbooks and exam PDFs, then run a limited produce batch first. I built it to explore how retrieval can be guided by a learner’s real assessment material rather than by a generic summary prompt.
Top comments (0)