StudyLoop

StudyLoop is a local-first study companion for people who learn better by doing, explaining, and being asked the next useful question. It pairs a browser workspace with AI mentors such as Kiro, keeps track of where you left off, and turns real practice into evidence you can revisit.

www.studyloop.dev — what StudyLoop is, who it is for, and what it honestly cannot do yet.

A real StudyLoop session with live Kiro responses; the topic and learner reply were supplied when recording.

What studying with it feels like

• Study Session gives you a live mentor that asks short Socratic questions, adapts to your energy, and leaves space for you to reason.
• Body Double gives you a calm workspace, a visible timer, and an agent beside you while you work through one task.
• Today, flashcards, and quizzes help you return to useful work without having to reconstruct the whole plan in your head.
• Sessions and teach-backs preserve what you actually practised. Notes can add context, but they are optional and are not treated as proof that something was learned.

Study Session	Body Double

StudyLoop was built by a neurodivergent learner moving from networking into data engineering. Its AuDHD support is part of the workflow: smaller starting steps, energy-aware pacing, visible closure, a parking lot for tangents, and prompts that challenge without shaming.

Is it for me?

You do not need to be a software developer to use the study workspace. You do need someone comfortable with a terminal for the current source installation, plus at least one supported AI command-line agent. If that setup is unfamiliar, the step-by-step setup guide is the best place to start—and reports about confusing instructions are genuinely useful contributions.

The current release is best suited to:

• self-directed learners who want an active study partner rather than an answer bot;
• AuDHD learners who benefit from low-friction starts and explicit session endings;
• technical learners using Kiro CLI, Codex, Claude Code, OpenCode, or pi;
• contributors interested in learning tools, accessibility, documentation, or Python.

Try one study session

StudyLoop currently installs from source on macOS or Linux and requires Python 3.12 or newer.

git clone https://github.com/NetDevAutomate/StudyLoop.git studyloop
cd studyloop
./scripts/install.sh
studyloop setup
studyloop doctor --fix
studyloop web

Open the URL printed in the terminal, choose Study Session, enter a topic, select your installed agent, and start. The browser workspace runs locally; your study database, plans, and generated materials stay on your machine unless you choose to sync or export them. The selected AI agent may still send conversation content to its own model provider; its privacy and billing terms apply.

Prefer the terminal? Start the same kind of session with:

studyloop study "Python decorators" --energy 6

Follow Your First Week for a gentle path through study, review, and session history.

For the session-memory reliability candidate, see conversation memory, repair and sync. The guide covers recovering missing local transcripts and the limits of sharing a database; release acceptance is still pending.

What is ready—and what is not

StudyLoop is an early open-source release, so the boundaries are worth making clear:

• installation is from a Git checkout; there is no current PyPI or Homebrew release;
• the Web UI is built for laptop and tablet layouts. Phone widths get a usable bottom tab bar rather than a broken one, but a phone is not a supported layout and no journey is tested at that size;
• the Web UI needs the local StudyLoop server and does not work offline;
• voice uses a configured Kokoro-compatible server, then falls back to operating system voices when available;
• study plans can be created in the Web UI or CLI, but the current Web UI form is manual—an agent-led planning interview is not integrated there yet;
• practice-task generation and verification are currently CLI workflows.

Those limits are tracked openly in the roadmap. If one blocks you, an issue describing the real workflow is more helpful than a feature wishlist without context.

Help shape StudyLoop

Contributions do not have to be code. Clear bug reports, setup notes, accessibility feedback, screenshots, documentation fixes, and descriptions of where a study flow became overwhelming are all welcome.

For code changes, small pull requests with a focused test are easiest to review. The contributing guide covers the development setup, checks, and pull request process.

• Open an issue
• Browse pull requests
• Read the contribution guide

Guides

The project website, www.studyloop.dev, is the shortest way to decide whether StudyLoop suits you. The guides below are the detail.

• Setup Guide — install and configure the current release
• Your First Week — reach a useful first routine gradually
• Web UI Guide — Study Session, Body Double, review, and access
• Study Plans — create and use plans without over-planning
• Content Pipeline — make local cards and practice material
• Troubleshooting — recover from common setup problems
• CLI Reference — complete command details
• AuDHD Learning Philosophy — why the workflow is designed this way

License and acknowledgements

MIT — see LICENSE.

StudyLoop's study-plan shape (mission first, learning records kept like decision records, primary sources over recalled knowledge) is adapted from Matt Pocock's teach skill, published under the MIT License, Copyright (c) 2026 Matt Pocock. The ideas were re-implemented rather than copied; his copyright notice and the MIT permission notice are kept in THIRD-PARTY-NOTICES.md so they travel with any portion that is ever reused verbatim.

Vendored fonts and JavaScript libraries, with their licences, are listed in the same file.