repowiserepowise

App

amosblomqvist / learnharness

Agent harness for learning difficult things.

★ 10,065stars on GitHubTypeScriptAGPL-3.0

This connects Repowise to amosblomqvist/learnharness's code. It does not install learnharness.

Repowise maps public repos from their source. Not affiliated with amosblomqvist. What is Repowise?

github.com/amosblomqvist/learnharnessIndexed at cb25c8e

What is learnharness?

LearnHarness is an app for learning difficult subjects, using AI agents to plan and teach lessons from your current understanding to a goal you set. The agents first measure where your understanding ends, then build a dependency map from that point to the goal and teach each step on top of what you already know. It is written mainly in TypeScript and runs as a desktop app on macOS, Windows and Linux. To install it, run git clone https://github.com/amosblomqvist/learnharness.git and follow the steps below. Its code health is 7.8 out of 10 (Good), ranked #82 of 4,000+ indexed repos. The last commit to learnharness was on Nov 12, 2026, and 42 people have contributed.

As of commit cb25c8e, indexed Nov 13, 2026. What it is comes from a model reading the README; the rest is measured from the source and git history, not from the maintainer.

For agentslearnharness.md

Repo facts: type, license, health, rank, contributors
learnharness facts
TypeApp
LicenseAGPL-3.0
Code health7.8 / 10 (Good)
Health rank#82 of 4,000+ indexed repos
Last commitNov 12, 2026
Contributors42
Bus factor1
Size1,169 files, 162,893 lines

Install learnharness

Commands copied from the README

  1. 1Clone the learnharness repository.
    git clone https://github.com/amosblomqvist/learnharness.git
  2. 2Move into the learnharness folder.
    cd learnharness
  3. 3Install the dependencies
    npm ci
  4. 4Set up the plugins
    npm run plugins:setup
  5. 5Start the development server
    npm run dev
  6. 6Run the tests
    npm test

Full install notes on GitHub

Explore learnharness

Ask about this code in your agent

This connects Repowise to learnharness's code. It does not install learnharness. To install learnharness, use the install steps.

Useful if you want to change or fork learnharness: your agent can ask how it is put together and get answers that cite its files.

Ask about learnharness

Answers come from learnharness's README and source, with the files they cite. Two free answers a day without an account.

Featured answer

learnharness is an app for learning difficult subjects, where AI agents first measure your current understanding and then plan and teach lessons toward a goal you set. It is written mainly in TypeScript and runs as a desktop app on macOS, Windows and Linux. The Electron entry point electron/main.cjs opens the desktop window, runs the local server, and holds sign-in state for ChatGPT and subscription accounts. The app directory defines the screens, components holds the React interface with components/chat.tsx running the session, and the agent logic sits under src/agent, including src/agent/subagents/extraction.ts for lesson material. The root config.ts sets where data is kept, the plugins directory adds features such as flashcards and a drawing pad, and the layers by file count are UI with 114 files, Application with 210, Service with 72 and Docs & Tooling with 170, out of 1169 files in total. To start reading, open electron/main.cjs for the desktop shell, components/chat.tsx for the session, and src/agent for the teaching logic.

Files per layer in learnharness9 layers.

Basis: Inferred · knowledge graph· as of cb25c8e

Want this for your own codebase? Map your repo free

About this page

Get a page like this for your own codebase

Repowise built this map of amosblomqvist/learnharness from its source and git history. It parses every symbol, draws the dependency graph, scores each file's health from complexity, duplication, coverage and churn, finds hotspots and owners in the history, and writes documentation for every module. Your coding agent reads the same index over MCP, so its answers cite real files.

Free for 2 public repos, with a page like this one. No card. See how Repowise works

Prefer local? pip install repowise && repowise init indexes your own checkout, no account needed. Star Repowise on GitHub

amosblomqvist/learnharness FAQ

What is LearnHarness?

LearnHarness is an app that uses AI agents to plan and teach lessons from your current understanding to a goal you set. It runs as a desktop app on macOS, Windows and Linux, and developers can run it from source.

How does LearnHarness work?

LearnHarness measures where your understanding ends, builds a dependency map to your goal, then teaches each step and offers practice problems. Your answers and thinking steer what comes next, and a knowledge graph carries what you learn across sessions.

How do I install LearnHarness?

Developers can install LearnHarness from source by running npm ci, then npm run plugins:setup, then npm run dev to start it. Learners can download the desktop app for macOS, Windows or Linux. See the install steps

What language is LearnHarness written in?

LearnHarness is written mainly in TypeScript, with some JavaScript and Python files. Its interface is built with React and its desktop shell uses Electron.

How is the LearnHarness code organized?

The code is split into app screens, React components, an Electron desktop shell and plugins, each plugin with its own entry point. Agent logic lives under src/agent, and the root config.ts holds shared paths and types.

Who is LearnHarness for?

LearnHarness is for learning difficult things, starting from what you already understand and working toward a goal you set. It asks you to explain your reasoning so the lesson can adapt to it.

What license does LearnHarness use?

LearnHarness is open source under the AGPL-3.0 license. The license id in the project's metadata is AGPL-3.0.

Code health

7.7out of 10Good

Hotspots average 6.6, the place fixes pay off most.

On the leaderboards:#33 of 70 TypeScript repos

Key features of learnharness

  • Measures where your understanding ends before teaching, then plans a dependency map from that point to your goal.
  • Lets you explain your reasoning, uncertainties and reactions in your own words by typing or using dictation, so the agent builds a detailed picture of what you understand.
  • Keeps a persistent knowledge graph of claims and the premises behind them, with links back to the source sessions, and lets later sessions read it.
  • Accepts files such as PDFs and photos in the desktop app, with the agent reading each file from its place on disk.
  • Stores chats, the graph, settings and API keys on your machine as plain files you can inspect, and works with provider keys, subscriptions, an installed Claude Code CLI or a custom OpenAI-compatible endpoint.

How amosblomqvist/learnharness works

The Electron entry point electron/main.cjs opens the desktop window, runs the local server and holds sign-in state for ChatGPT and subscription accounts. The app directory defines the screens, including the home page, the new session page, the chat and the settings. The components directory holds the React interface, and components/chat.tsx runs the session while drawing the transcript from shared widgets. The agent logic lives under src/agent, where src/agent/subagents/extraction.ts works on extracting material for the lesson. The plugins directory contains feature projects such as flashcards and a drawing pad, each with its own src/main.ts or src/main.js entry point. The root config.ts sets where the app keeps its data and provides the path table and thinking-level type that src/agent, src/store, src/server and src/core import.

The codebase has 1,169 files and 7,329 symbols in 12 modules, led by TypeScript, JavaScript and Python. Code health is 7.7 out of 10, rated good.