Study set generator from uploaded course materials
Build a workflow called Studycraft that turns a student's course material into a complete study set. A student uploads any mix of documents — PDFs, Word docs, PowerPoint slides, Excel spreadsheets, images, plain notes, or a whole folder — and gets back flashcards of the key information, a full study guide, and two interactive tests (an easy one and a hard one), each mixing multiple choice, true/false, and written short/long answers. The workflow is browser-triggered via HTTP routes, gated so only authenticated members of the organization can use it, and every study set is private to the person who created it (scoped to their authenticated email). Persist study sets in a named volume so they survive across runs and each set is owned by its creator. Design it as a queue-and-poll system, because full generation can take a couple of minutes and would otherwise blow past the HTTP gateway timeout: A web app page that shows a library of the user's saved sets, an uploader for files or a folder, and a viewer for opening a completed set. Build this as a polished, production-grade UI. An upload/generate endpoint that immediately queues the job — stashing the raw upload and a "pending" set into storage — and returns right away so the browser never waits on the model. A background worker (linked, not routed) that extracts text from Office/text/spreadsheet files, hands PDFs and images natively to the AI model, and has the model produce the flashcards, study guide, and both tests, then rewrites the stored set as "ready" (or "error"). Use Anthropic Claude for generation — or whichever AI model provider you prefer — via a connector. A sets endpoint the app polls until a set is ready. A grading endpoint that scores a submitted test — instantly for multiple-choice/true-false, and using the AI model for written answers — returning a score and per-question feedback. Keep concurrent generations from conflicting by giving each set its own file in storage. Don't modify any external systems beyond reading/writing the study-set storage. Tools used: Anthropic Claude API, 3B
What this prompt builds
Studycraft transforms uploaded course materials (PDFs, documents, images, folders) into comprehensive study sets including flashcards, a study guide, and two interactive tests with multiple question types. Built as a queue-and-poll system to handle long AI generation times, it persists every user's study sets privately in named volumes and uses Claude to extract key information and grade written answers. The workflow provides a polished web interface with a personal library, uploader, and test-taking experience, all gated to authenticated organization members.
The problem
Students who need to turn raw course materials into study aids face slow, limited tools — Quizlet's AI often misses key information and caps uploads behind paywalls. Studycraft solves this by accepting unlimited uploads of any document type (PDFs, Word files, PowerPoints, images, folders) and automatically generating a complete study package: flashcards of the essential facts, a full study guide, and two interactive tests (easy and hard) mixing multiple-choice, true/false, and written questions. Answers stay hidden until the test is submitted, and every study set is saved privately in the user's personal library for repeated review.
Solution and impact
Studycraft eliminates the manual grind of creating study materials and removes upload limits or paywalls. Students upload once and immediately receive a ready-to-use study set, with AI-powered grading that scores multiple-choice instantly and evaluates written answers for correctness and completeness. By persisting every set in private storage and running generation in the background, the workflow delivers a fast, polished experience that scales to any number of courses or exam cycles without external dependencies.