OpenQuest

From Iron to Oracle - contribute your way through the open source world and work your way to the top! Brush up on your programming skills using our AI agent which can guide you through your journey.

Comment

Inspiration

As junior developers, we all want to contribute to open source, but we often don't know where to start. Facing thousands of GitHub repositories and issues, we feel lost, intimidated, and confused—unsure which projects match our skill level, afraid our code isn't good enough, and overwhelmed by massive codebases.

We wanted to build "LeetCode for Open Source"—a platform where anyone can find issues that match their abilities and get AI guidance to land their first PR.

What it does

OpenQuest is a gamified open-source contribution platform:

  • Smart Matching: AI recommends projects based on your tech stack, experience level, and interests
  • Issue Ranking: Automatically analyzes issue difficulty to help you find true "good first issues"
  • AI Mentor: An Agent explores the codebase, explains root causes, and guides you step-by-step
  • Sandbox Execution: Run code safely in isolated cloud environments without breaking your local setup
  • Gamification: Earn XP for merged PRs, level up to unlock harder projects, and collect badges

How we built it

  • Frontend: React + TypeScript + Tailwind CSS with a terminal-style UI that shows Agent analysis in real-time
  • Backend: FastAPI + Python implementing an Agent Loop where AI autonomously decides which files to read and what code to search
  • E2B Sandbox: Isolated cloud VMs for safe Git operations and code analysis
  • Claude API: Powerful LLM for code understanding and problem analysis
  • GitHub OAuth: Securely push code to users' forks
  • SSE Communication: Real-time streaming of Agent's thinking process to the frontend

Challenges we ran into

  • Context Management: LLM context grows with each turn—implemented truncation strategies and sliding windows
  • Large Repo Handling: Cloning repos like React took forever—optimized with shallow clones
  • Real-time Sync: WebSocket connections kept dropping—switched to SSE for stability
  • GitHub Permissions: Properly syncing forks with upstream so PRs can be correctly compared

Accomplishments that we're proud of

  • Built a complete AI Agent loop that autonomously explores codebases, analyzes issues, and generates solutions
  • Created a terminal-style real-time UI showing every step of the Agent's thinking
  • Implemented end-to-end automation from analysis to pushing code
  • Made open-source contribution fun and approachable through gamification

What we learned

  • Agentic AI design patterns: How to let LLMs make decisions and execute tool calls autonomously
  • Sandbox Technology: How E2B spins up isolated execution environments in milliseconds
  • Real-time Communication: Trade-offs between SSE and WebSocket
  • Open Source Community: Truly understanding the pain points of first-time contributors

What's next for OpenQuest

  • Auto-complete code fixes: Agent not only analyzes but generates the actual fix
  • PR Review Assistant: Check code quality before you submit
  • Community Leaderboard: Contributors motivate each other through friendly competition
  • IDE Plugin: Use OpenQuest directly in VSCode/Cursor

Built With

Share this project:

Updates