Sandbox
@hluaguo/learn-faster-kit

CLI learning coach for Claude Code and Codex

Learn FASTER is a project-level CLI that adds a learning workflow around Claude Code and Codex. It creates shared `.learning/` files, agent instructions, and Claude Code slash commands for `/learn`, `/review`, and `/progress`, then uses those pieces to generate syllabi, schedule reviews, and track progress.

374 stars39 forksPythonUpdated 1mo ago
Who it's for

Builders who want their agent to help them learn a topic or codebase with guided practice and spaced repetition.

What it delivers

You can turn an agent session into a structured study loop instead of re-explaining what you want to learn each time.

What it does

Personalized syllabi

Generates a syllabus based on your skill level, goals, and chosen learning mode.

Spaced repetition reviews

Schedules reviews for topics you have learned and surfaces them again at the right time.

Learning modes

Offers Balanced, Exam-Prep, Theory-Focused, and Practical modes for different study styles.

Active practice prompts

Creates exercises and projects so you practice by doing, not only by reading.

Progress tracking

Stores learning progress and shows session history and statistics.

Agent-specific setup

Installs Claude Code commands and instructions, and can write Codex instructions to AGENTS.md.

How to get it

  1. 1Prerequisites: uv package manager
    # Install uv if you haven't already
    curl -LsSf https://astral.sh/uv/install.sh | sh
  2. 2Install once and use across all projects
    uv tool install learn-faster --from git+https://github.com/cheukyin175/learn-faster-kit.git
  3. 3Then in any project directory, simply run
    learn-faster
  4. 4Run directly without installation
    uvx --from git+https://github.com/cheukyin175/learn-faster-kit.git learn-faster
  5. 5Install the tool
    uv tool install learn-faster --from git+https://github.com/cheukyin175/learn-faster-kit.git
  6. 6Launch in any project directory
    cd your-learning-project
    learn-faster

README

Learn FASTER

Python Version License: MIT uv

AI-powered learning coach that accelerates mastery through spaced repetition, personalized syllabi, and active practice.

Built for Claude Code - Integrates AI coaching directly into your development environment. Also supports Codex with agent-specific AGENTS.md instructions and shared learning tools. Codex support is less strict than Claude Code because Codex does not currently expose a CLI flag for replacing the system prompt; Learn FASTER can only inject coaching behavior through AGENTS.md, startup prompts, or future skill-style instructions. Codex also only exposes its structured user-input tool in Plan mode, so in Default mode it cannot present guided choice prompts the same way Claude Code can.

Why Learn FASTER?

Master any technical skill with science-backed learning principles:

  • Personalized syllabi generated for your skill level and learning goals
  • Spaced repetition system that schedules reviews at optimal intervals
  • Four learning modes - choose Balanced, Exam-Prep, Theory-Focused, or Practical
  • Active practice with auto-generated exercises and projects
  • Progress tracking to visualize your learning journey

The FASTER Framework

  • Forget: Beginner's mindset - approach topics with fresh perspective
  • Act: Learn by doing - hands-on practice over passive reading
  • State: Optimize focus - create ideal learning conditions
  • Teach: Explain to retain - teaching reinforces understanding
  • Enter: Consistent sessions - regular practice builds momentum
  • Review: Spaced repetition - review at intervals for long-term retention

Installation

Prerequisites: uv package manager

# Install uv if you haven't already
curl -LsSf https://astral.sh/uv/install.sh | sh

Option 1: Persistent Installation (Recommended)

Install once and use across all projects:

uv tool install learn-faster --from git+https://github.com/cheukyin175/learn-faster-kit.git

Then in any project directory, simply run:

learn-faster

This will auto-initialize on first run and launch Claude Code with FASTER coaching mode.

Option 2: One-Time Use

Run directly without installation:

uvx --from git+https://github.com/cheukyin175/learn-faster-kit.git learn-faster

What Gets Installed

On first run, learn-faster creates shared learning tools plus agent-specific instructions.

For Claude Code:

your-project/
├── .claude/
│   ├── agents/practice-creator.md
│   ├── commands/
│   │   ├── learn.md
│   │   ├── review.md
│   │   └── progress.md
│   └── settings.local.json
├── .learning/
│   ├── config.json (tracks initialization)
│   ├── scripts/
│   │   ├── init_learning.py
│   │   ├── log_progress.py
│   │   ├── review_scheduler.py
│   │   └── generate_syllabus.py
│   └── references/faster_framework.md
└── CLAUDE.md

For Codex, run learn-faster init --agent codex. It creates the same .learning/ shared tools and writes Codex instructions to AGENTS.md. Because Codex does not provide a system-prompt replacement flag, adherence depends on Codex reading those project instructions and the launch prompt. Codex guided-choice interactions are also limited: the user-input tool is available only in Plan mode, so Default mode falls back to plain conversational questions.

Quick Start

  1. Install the tool

    uv tool install learn-faster --from git+https://github.com/cheukyin175/learn-faster-kit.git
    
  2. Launch in any project directory

    cd your-learning-project
    learn-faster
    

    First run will:

    • Prompt you to select a learning mode
    • Initialize the project structure
    • Launch your configured agent with FASTER coaching enabled
  3. Start learning

    /learn "Golang fundamentals"
    

    The AI coach will generate a personalized syllabus and guide your learning session.

Demo: Teach-Back in Action

The "T" in FASTER—teaching to retain—is the key differentiator. Here's how it works:

mkdir learn-go && cd learn-go
learn-faster                    # Select "Balanced" mode
/learn "Go error handling"      # In Claude Code
Coach: You've just learned about error wrapping. Ready to teach it back?
       ┌ Teach Back
       │ ● Yes, let me explain
       │ ○ Need review first
       │ ○ Not sure yet
       └

You:   So when you wrap an error with fmt.Errorf and %w, you're adding
       context like "failed to open config" while keeping the original
       error inside. Then errors.Is can still match the root cause.

Coach: ✅ Great explanation! You nailed the key insight—wrapped errors
       preserve the chain for inspection. Adding "error wrapping" to
       your review schedule. First review tomorrow.

Why this works: Explaining concepts in your own words forces active recall—proven to boost retention 2-3x vs passive reading. The coach won't just tell you answers; it guides you to construct understanding yourself.

Usage

CLI Commands

  • learn-faster - Launch Claude Code with FASTER coaching (auto-initializes on first run)
  • learn-faster init - Force re-initialization or switch learning modes
  • learn-faster init --agent codex - Initialize the project for Codex instead of Claude Code
  • learn-faster resume [<id>] [--pick] [--fork] - Resume a previous coaching session (--pick to choose interactively; --fork to branch into a new session id)
  • learn-faster version - Show current version

Claude Code Slash Commands

Once Claude Code is running, use these commands:

  • /learn [topic] - Start or continue learning a topic with personalized syllabus
  • /review - Spaced repetition review session for topics you've learned
  • /progress - View detailed progress report and learning statistics

Learning Modes

Choose the mode that fits your learning style:

  • Balanced - Mix of theory, practice, and real-world application (recommended for most learners)
  • Exam-Prep - Focused on recall, practice tests, and certification preparation
  • Theory-Focused - Deep conceptual understanding with mental models and first principles
  • Practical - Project-based learning with immediate application

Each mode provides a tailored coaching experience with mode-specific syllabi and exercises.

Development

Setup

# Clone the repository
git clone https://github.com/cheukyin175/learn-faster-kit.git
cd learn-faster-kit

# Install uv if needed
curl -LsSf https://astral.sh/uv/install.sh | sh

# Install dependencies
uv sync

Use Cases

Learn FASTER is ideal for:

  • Learning new programming languages (Go, Rust, Python, TypeScript, etc.)
  • Preparing for technical certifications and exams
  • Mastering frameworks and libraries (React, Next.js, Django, etc.)
  • Building structured self-study programs
  • Onboarding to new codebases or technologies

Requirements

Contributing

Contributions are welcome! Feel free to open issues or submit pull requests.

License

MIT License - See LICENSE file for details

Files in the repo

Repository payload10 top-level entries
  • .github
  • src
  • tests
  • .gitignore
  • AGENTS.md
  • CHANGELOG.md
  • LICENSE
  • pyproject.toml
  • README.md
  • uv.lock

Discussion (0)

Ask about usage, or say what you built with it

Sign in to join the discussion.

No comments yet. Be the first to say what this is good for.

More tools

JuliusBrussee/
caveman

🪨 why use many token when few token do trick — Claude Code skill that cuts 65% of tokens by talking like caveman

105k
1 add
MemPalace/
mempalace

The best-benchmarked open-source AI memory system. And it's free.

59k
stablyai/
orca

Orca is the ADE for working with a fleet of parallel agents. Run any coding agent with your own subscription. Available on desktop, mobile and remote runtime.

66k

A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.io

132k

Never stop coding. Free MIT AI gateway: one endpoint, 352 providers (150+ free), 1200+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cline & Copilot. Quota-aware auto-fallback, RTK+Caveman compression saves 15-95% tokens, MCP/A2A, Desktop/PWA. Built by 550+ contributors

64k
headroomlabs-ai/
headroom

Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.

71k