Skip to content

Latest commit

Β 

History

49 Commits

Folders and files

NameName
Last commit message
Last commit date
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 

Repository files navigation

LemonBeam πŸ‹

LemonBeam shines a fresh beam of light on an unfamiliar codebase.

It scans a public JavaScript or TypeScript GitHub repository and generates a fixed-format contributor guide that helps new developers understand the project more quickly. Rather than sending an entire repository directly to an LLM, LemonBeam classifies repository files, creates meaningful chunks, retrieves only the evidence relevant to each guide section, and generates a source-backed guide with citations.


Why LemonBeam?

Understanding an unfamiliar repository is difficult.

Important information is often scattered across:

  • README files
  • package.json scripts
  • configuration files
  • folder structure
  • test suites
  • source code

Developers often paste a repository into an AI assistant hoping for an explanation, but the results depend heavily on the prompt and available context.

LemonBeam takes a deterministic approach by:

  • scanning the repository
  • organizing repository evidence
  • retrieving only the information relevant to each guide section
  • generating a repeatable, source-backed contributor guide

Features

MVP

  • Scan public GitHub JavaScript and TypeScript repositories
  • Generate a fixed-format contributor guide
  • Source-backed sections with citations
  • Tree-sitter parsing for JavaScript and TypeScript
  • Rule-based repository classification
  • Deterministic SQLite retrieval
  • React web interface

Example Guide Structure

The generated guide includes:

  • Repository Information
  • Project Overview
  • Prerequisites
  • Installation and Environment Setup
  • Running and Building
  • Project Structure
  • Testing
  • Key Files and Entry Points
  • Suggested Reading Order
  • Uncertainties and Missing Information
  • Source Citations

Technology Stack

Frontend

  • React
  • TypeScript
  • Vite
  • Tailwind CSS

Backend

  • Node.js
  • Express
  • TypeScript
  • SQLite
  • Tree-sitter
  • GitHub API
  • OpenRouter API (BYOK)

Repository Structure

frontend/
backend/

README.md
PROJECT_BRIEF.md
ARCHITECTURE.md
DATABASE.md
API_CONTRACT.md
TESTING.md
DECISIONS.md
CONTRIBUTING.md
AGENTS.md

Getting Started

Prerequisites

  • Node.js
  • npm

Installation

git clone <repository-url>

cd lemonbeam/frontend
npm install

cd ../backend
npm install

Using the CLI

LemonBeam can be run directly from your terminal as a command-line tool to analyze any local project directory and generate an AI-powered documentation guide.

1. Configure Your Environment

Create a .env file in the root of your project directory and add your OpenRouter API key:

OPENROUTER_API_KEY=your_openrouter_api_key_here

2. Link LemonBeam Locally

If you are testing or running the tool locally from source, link it to your project:

npm link lemonbeam

3. Run the CLI

Navigate into the root of any project you want to scan and execute:

npx lemonbeam

4. Choose Your Output Format

When the scanning pipeline finishes, you'll be prompted interactively in your terminal:

  • Type y to automatically save the generated guide as a markdown file in your project directory.
  • Type n to print and view the generated guide right in your terminal.

Running the Project

Frontend

cd frontend

npm run dev

Backend

cd backend

npm run dev

Running Tests

To run the automated test suite, use the following commands:

# Run tests once
npm test

# Run tests in watch mode
npm run test:watch

Documentation

PROJECT_BRIEF.md β€” project goals, MVP, user flow, and technical challenges ARCHITECTURE.md β€” system architecture and backend design DATABASE.md β€” temporary SQLite schema and relationships API_CONTRACT.md β€” frontend/backend API specification TESTING.md β€” testing strategy DECISIONS.md β€” architectural decisions and rationale CONTRIBUTING.md β€” contributor workflow AGENTS.md β€” instructions for AI coding agents

Contributing

Please read CONTRIBUTING.md before opening an issue or submitting a pull request.


License

This project is currently under development. License information will be added before the first public release.

About

A tool that shines a beam of light on an unfamiliar codebase and refracts it into a clear, reliable guide.

Resources

Code of conduct

Contributing

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages