DevForge AI – AI Engineering Workspace powered by Gemini
Inspiration
Developers spend a significant amount of time understanding existing codebases before they can contribute effectively. Whether onboarding to a new team, exploring open-source projects, or maintaining legacy software, navigating thousands of lines of code with limited documentation is time-consuming and frustrating.
We wanted to build a tool that transforms any GitHub repository into an intelligent engineering workspace, allowing developers to understand project architecture, explore code, generate documentation, and receive AI-powered insights—all from a single platform.
What it does
DevForge AI analyzes any public GitHub repository using Google Gemini and provides developers with an interactive engineering workspace.
Key capabilities include:
- 🤖 AI-powered repository chat
- 🏗️ Architecture visualization
- 📄 Automatic documentation generation
- 💡 Code explanation and repository insights
- 🛡️ Bug and security analysis
- 📊 Engineering health reports
- 📑 Exportable project documentation
Instead of manually reading hundreds of files, developers can simply ask questions about the repository and receive contextual, repository-aware answers.
How we built it
We built DevForge AI using a modern full-stack architecture.
- Frontend: Next.js, React, TypeScript, Tailwind CSS
- Backend: FastAPI (Python)
- AI: Google Gemini 2.5 Pro
- Authentication: Firebase Authentication
- Database: Firestore
- Git Integration: GitHub API
Our workflow is:
- User connects a GitHub repository.
- The backend clones and parses the repository.
- Project structure, dependencies, and code summaries are extracted.
- Structured repository context is sent to Gemini.
- Gemini generates documentation, architecture insights, explanations, and engineering recommendations.
- Results are displayed through an interactive dashboard.
Challenges we ran into
One of our biggest challenges was handling large repositories without exceeding AI context limits. Instead of sending the entire codebase, we designed a parsing pipeline that extracts only the most relevant structural information before passing it to Gemini.
Another challenge was presenting complex engineering information in a simple and intuitive way. We focused on creating visual architecture diagrams and concise AI-generated summaries to improve the developer experience.
Accomplishments that we're proud of
- 🚀 Built an AI Engineering Workspace instead of just another coding assistant.
- 🧠 Successfully integrated Google Gemini for repository-aware reasoning.
- 🏗️ Automated architecture understanding from GitHub repositories.
- 📄 Generated professional project documentation with minimal user input.
- 💬 Enabled developers to ask natural-language questions about an entire codebase.
- 📊 Combined multiple developer tools into one unified platform.
What we learned
This project taught us how repository parsing, structured context generation, and large language models can work together to solve real software engineering challenges.
We also learned that providing AI with structured repository information dramatically improves response quality compared to sending raw source code. Most importantly, we gained hands-on experience building scalable AI-powered developer tools using Google Gemini.
What's next for DevForge AI
We're excited to continue expanding DevForge AI with features including:
- 🔄 Multi-repository workspaces
- 👥 Team collaboration
- 🔍 Pull Request reviews
- 💻 VS Code extension
- ⚙️ CI/CD integration
- ☁️ Docker and Kubernetes analysis
- 📈 Repository health scoring
- 🔔 GitHub webhook automation
- 🤝 Jira and Linear integration
Our long-term vision is to make DevForge AI the go-to AI Engineering Workspace that helps developers understand, maintain, and improve software projects faster than ever.
Log in or sign up for Devpost to join the conversation.