An AI-powered GitHub profile analyzer that delivers brutal, honest assessments of developer code quality, behavioral patterns, and career trajectory.
Try out the GitDeep Live Application directly in your browser.
GitDeep is a client-side web application using advanced AI models to evaluate GitHub profiles with behavioral analysis, buzzword detection, career trajectory tracking, and per-repository assessments.
- Dual Assessment Modes: Employer Mode (hirability analysis and scoring) and Developer Mode (mentorship and actionable guidance).
- Deep Analysis: Career slope detection, buzzword vs reality verification, AI usage quality assessment, behavioral pattern detection, and per-repository scoring.
- Privacy-Focused Architecture: Session-based execution with zero server database storage, telemetry, or tracking. Gemini and Anthropic calls are executed directly client-side, while requests to Ollama and OpenAI-compatible providers pass through a stateless Next.js API proxy (
app/api/ai/route.ts) to avoid CORS restrictions without persisting API keys or data. - AI Provider Flexibility: Supports 12+ cloud and local AI providers including Google Gemini, OpenAI, Anthropic, Groq, DeepSeek, OpenRouter, and local Ollama instances.
- Rich Visualizations: Interactive radar charts, career timeline meters, language distributions, and comparative candidate analysis.
- Node.js 18.18.0+ and npm
- An AI provider API key (Google Gemini recommended for free tier)
git clone https://github.com/Yuvraj-Sarathe/GitDeep.git
cd GitDeep
npm install
npm run devAccess the local development server at GitDeep Local Host.
- Visit Google AI Studio API Key Setup to obtain a free API key.
- Open GitDeep Settings (gear icon in the navigation bar).
- Select your AI Provider (e.g., Gemini API), enter your key, and select your target model.
- (Optional) Provide a GitHub Personal Access Token (PAT) under Settings to increase GitHub API rate limits (from 60 to 5,000 requests/hour).
Choose a model based on hardware and assessment requirements:
| Provider | Recommended Model | Prompt Size | Employer Mode | Notes |
|---|---|---|---|---|
| Gemini 2.5 Flash | Full | Yes | Recommended default (free tier available) | |
| OpenAI | GPT-4o | Full | Yes | High accuracy and reasoning |
| Anthropic | Claude Sonnet 4 | Full | Yes | Excellent analysis quality |
| Ollama | Qwen 2.5 7B | Full | Yes | Recommended for local 8GB RAM systems |
| Ollama | Llama 3.1 8B | Full | Yes | Recommended for local 16GB RAM systems |
For detailed model benchmarks, small model optimization, and local setup steps, refer to the Ollama & Local Models Guide.
Build and run GitDeep in a container:
docker build -t gitdeep .
docker run -p 3000:3000 gitdeepOpen the application at GitDeep Local Host Container.
GitDeep uses Next.js 15, TypeScript, Tailwind CSS, and Octokit:
- Data Collection: Client fetches public profile data, non-fork repository metadata, README snippets, and merged PRs via the GitHub REST API.
- AI Processing: Gemini and Anthropic requests are called directly from the browser; Ollama and OpenAI-compatible requests are proxied via the Next.js API route (
app/api/ai/route.ts) to manage headers and avoid CORS issues. - Visualization: Formats structured JSON responses into hirability scores, SWOT matrices, radar charts, and mentorship steps.
For a full technical overview and directory layout, see Architecture Documentation.
- Small Models: Local models under 7B parameters may yield incomplete or shallow responses.
- GitHub Rate Limits: Unauthenticated GitHub API calls are limited to 60 requests per hour; using a Personal Access Token increases this to 5,000 requests per hour.
For a detailed list of constraints and upcoming features, see Limitations & Roadmap.
Contributions are welcome. Please read CONTRIBUTING.md for development environment setup, code style standards, and pull request procedures.
This project is licensed under the GNU General Public License v3.0 - see the LICENSE file for details.
Yuvraj Sarathe
Yuvraj-Sarathe |
MILAN-123865 |
PrathamReddy888 |
github-actions[bot] |
M0izz |
karrisanthoshigayatri |
preranaanand07 |
- GitHub API for data access
- Google Gemini for free-tier AI capabilities
- Ollama for local model integration
- Next.js and Vercel for web infrastructure
- Open an issue on GitHub Issues
- Access the live application at GitDeep Live Application