CodeAtlas

Your Smartest Teammate- An Agent-Powered Project Manager that creates tasks from your project description, updates progress automatically, and visualizes the project in mere minutes.

Comment

Inspiration

Our goal was to make working on projects easier and more organized while also simplifying collaboration on team projects. We also wanted to lower the barrier to contributing to open source projects by making it clearer what tasks need to be worked on. Too often, there's a disconnect between writing code and seeing progress in a project. CodeAtlas helps by visualizing tasks in a structure that makes it easier for developers to stay on track and for contributors to get involved. For hackathons and other projects.

What it does

CodeAtlas connects to a project and visualizes it as a clear task board and tree. It organizes work into tasks and subtasks, helping developers see what needs to be done at a glance. This makes it especially helpful for open source projects where contributors can more easily find entry points and start contributing, while also showing a breakdown of the project by frontend, backend, database, and other components.

How we built it

We built a FastAPI server that receives GitHub webhooks and processes them through a Google Gemini AI agent. The system uses Google's Agent Development Kit (ADK) to create an intelligent classifier that:

  • Extracts project context from webhook data (repository, commits, PRs, issues)
  • Queries Supabase database to get current project tasks and their relationships
  • Analyzes semantic meaning of code changes, commit messages, and file modifications

Challenges we ran into

  • Server Communication: Our Next.js server had trouble communicating with our google ADK server.
  • Database Schema Design: Creating flexible hierarchical structures for different project types required careful planning. We implemented nodes, edges, and relations tables to support various project structures.
  • AI Prompting and Output: Prompting the agent correctly and ensuring it returned properly structured output was a challenge that required multiple iterations and testing.

Accomplishments that we're proud of

  • Created a system that automatically generates and visualizes tasks
  • Built flexible architecture that works with any GitHub repository and project structure

What we learned

  • AI Agent Development: How to build sophisticated agents with Google's ADK framework
  • Database Design: Creating hierarchical project structures with proper relationships
  • Real time Integration: Connecting external APIs to database updates seamlessly

What's next for CodeAtlas

  • Enhanced AI Capabilities: Integrate more sophisticated AI models (Cursor) for better context understanding and support for more programming languages and frameworks.
  • Analytics Dashboard: Build a comprehensive dashboard showing project progress, developer productivity metrics, and task completion trends.
  • Integration Expansion: Support for other version control systems (GitLab, Bitbucket) and project management tools (Jira, Linear, Notion).
  • Advanced Workflow Automation: Implement custom workflow rules, automated code reviews, and intelligent task prioritization based on project dependencies.
  • Commit Tracking Agent: Add an agent that monitors commits and automatically updates task status when work is completed.

Built With

Share this project:

Updates