RepoAnalysis

Local CLI usage

The fastest way to use RepoAnalysis is as a local CLI. It reads your LLM credentials from a .env file and writes a Markdown tutorial to disk.

1. Configure

Create backend/.env (or .env in the directory you run from):

LLM_PROVIDER=gemini
GEMINI_API_KEY=sk-...

Other supported providers: anthropic, openai, huggingface. For Hugging Face, set LLM_MODEL=Qwen/Qwen2.5-Coder-32B-Instruct (or any chat-capable repo). For private or large repos, also set GITHUB_TOKEN=ghp_....

2. Install

cd backend
pip install -e .

3. Run

repo-analyze "https://github.com/owner/repo"

The tutorial is written to ./out/owner-repo/:

out/
└── owner-repo/
    ├── index.md
    ├── 01-<chapter>.md
    ├── 02-<chapter>.md
    └── ...

Open index.md in any Markdown viewer that supports Mermaid (VS Code, Obsidian, GitHub, etc.) to navigate the chapters.

Options

repo-analyze <url> [--out DIR] [--max-abstractions N] [--language LANG] [--quiet]