Github-Roast

AI-powered GitHub developer profile analyzer — savage roasts, sharable reports, screenshot-worthy.

中文版

---

Overview

Github-Roast transforms any GitHub username or repo URL into a 13-dimension AI roast report — plus a custom formula that calculates your GitHub account's monetary value. Enter a GitHub handle → scrape their public profile, repos & events → DeepSeek AI generates a hilarious, brutally honest analysis covering roast, tech stack, weaknesses, career advice, and more.

Preview

Profile + Stats + Account Value (left) and Roast + Tech Stack + Weaknesses (right)

Features

• Scrape — fetches profile, repos (up to 100), and public events via GitHub REST API (no browser needed)
• Analyze — sends formatted data to DeepSeek V4 Flash (free) or V4 Pro, returns structured JSON
• 13 Report Cards — about, roast, tech stack, weaknesses, open source influence, project highlights, collaboration style, activity pulse, career advice, achievements, life suggestion
• Account Value — proprietary formula calculates your GitHub worth in ¥ or $ based on stars, followers, original repos, and forks
• Download as Image — section-by-section PNG export via html2canvas, plus "Download All"
• Share to X — pre-filled tweet with your report card
• Bilingual UI — Simplified Chinese / English toggle
• Supports both Users & Organizations — auto-detects account type, uses different analysis prompts

Quick Start

Try it online

👉 groast.streamlit.app

Run locally

# 1. Clone
git clone https://github.com/gokuscraper/github-roast.git
cd github-roast

# 2. Install Python deps
pip install -r requirements.txt

# 3. Configure API keys
# Create .streamlit/secrets.toml with:
# GITHUB_TOKEN = "ghp_your-github-token-here"
# OPENCODE_API_KEY = "sk-your-opencode-key-here"

# 4. Run
streamlit run streamlit_app.py

API Keys

Key	Required	Where to get
GITHUB_TOKEN	Yes	GitHub Tokens (public_repo scope)
OPENCODE_API_KEY	Yes	OpenCode free channel
SILICON_API_KEY	Optional	SiliconFlow Console — used as fallback

Project Structure

github-roast/
├── streamlit_app.py        # Home page — username/URL input → scrape
├── pages/
│   └── 1_Analysis.py       # 4-section analysis report page
├── lib/
│   ├── ai.py               # AI prompts (individual/org), dual API strategy
│   ├── repo_utils.py       # GitHub data formatting & stats
│   └── sidebar.py          # Sidebar navigation
├── github_scraper/
│   ├── __init__.py         # fetch_all() entry point
│   ├── client.py           # GitHub API client wrapper
│   ├── config.py           # API endpoints config
│   ├── fetcher.py          # Fetch user + repos + events
│   ├── models.py           # GitHubUser, GitHubRepo, GitHubEvent dataclasses
│   ├── parser.py           # Parse raw API data → models
│   └── storage.py          # Local JSON caching
├── locales/
│   ├── zh.json             # Chinese UI strings
│   └── en.json             # English UI strings
├── i18n.py                 # i18n helper
└── requirements.txt

How It Works

flowchart LR
    A[GitHub URL / Username] --> B[GitHub REST API]
    B --> C[Profile JSON + Repos + Events]
    C --> D[DeepSeek AI]
    D --> E[13-Field JSON Report]
    E --> F[4-Section Card Grid UI]
    F --> G[Download PNG / Share to X]

Loading

1. Input — GitHub username, full profile URL (https://github.com/user), or repo URL (https://github.com/user/repo)
2. Scrape — fetches public data via GitHub REST API (/users/{user}, /users/{user}/repos, /users/{user}/events/public)
3. Detect — auto-detects User vs Organization and picks the matching analysis prompt
4. Analyze — formatted data sent to DeepSeek with a savage, witty prompt in your chosen language
5. Report — 13 cards grouped into 4 downloadable sections
6. Share — download section-by-section or all-at-once as PNG, or share to X

Tech Stack

Layer	Technology
Frontend	Streamlit (single-page app)
AI Model	DeepSeek V4 Flash (free) / V4 Pro via OpenCode + SiliconFlow fallback
Data Source	GitHub REST API
Screenshot	html2canvas
Deployment	Streamlit Cloud
License	Apache 2.0

License

Apache 2.0