Project Info

DoGood

Inspiration

We live in an epidemic of burnout and disconnection and the online world effectively fills this gap with instant gratification through the dopamine of scrolling and easy rewards; our vision was to leverage these patterns to Do Good. Through trackable acts of kindness, productivity, and self-growth, DoGood guides users in building purpose while creating a tangible, real-world interest.

What it does

DoGood will present opportunities for social good and self-improvement through four pages that have tasks that require photo verification: Serve: The first tab, Serve, connects you with ways to give back. You can find local volunteer opportunities, community service projects, or even respond to crisis-assistance alerts. There’s also an environmental mini-game that turns recycling and trash pickup into friendly local competitions where you log what you clean, earn XP, and see your positive impact grow. Be Productive: The Be Productive tab helps you make good use of your time. You can set focus timers for tasks, and the app sends gentle notifications to keep you accountable. It also uses smart tracking to remind you of activities you haven’t done in a while, like that personal project you left half-finished, or that hobby you love but forgot. Completing tasks or having a productive work session with the timer can also earn XP. Self-improve: The Self-Improve tab remembers your progress and gives personalized suggestions to earn XP, maybe journaling today, maybe reconnecting with an old friend. Over time, it becomes your personal coach for growth, mindfulness, and relationships. Shop: Finally, all this effort pays off, literally for good causes. Every task completed earns XP. You can spend your XP in the Shop to fund real charitable donations, from supporting food banks to environmental nonprofits. You’re improving yourself and your community at the same time. If a user opens the app and isn’t sure what part of doing good they want to prioritize for the day, they can speak with the “DoGood companion,” that can work with the user to narrow down goals and provide direction to a page that aligns with what the user can accomplish.

How we built it

Once narrowing down the scope of what we wanted to accomplish for this version of DoGood, we began working on the frontend and UI/UX by describing the functionality and the vision for the four main buttons and debugging any faulty design through the site. We integrated this code to our GitHub repository with more frontend development made with Cursor and React. We finally integrated sponsors like Claude for the functionality and verification of all the features and LiveKit for the DoGood companion.

Challenges we ran into

The biggest challenges we ran into was narrowing down the scope for this version of the app because the inspiration to DoGood could mean a lot of things pertaining to social good and self-improvement with endless possibilities. Another challenge we ran into was debugging tools including LiveKit and Interaction Co to be integrated into the app.

Accomplishments we're proud of

Accomplishments that we are proud of are streamlining a clean UI/UX for different iterations of the app, leveraging Claude for verification, prompt engineering a LiveKit chatbox, and creating context API for knowledge transfer between different agents.

What's next

The next steps for DoGood are giving it an aspect of community–having challenges within schools like USC or between different schools or group affiliations to encourage competition for the dopamine of earning XP. With a community page, users can organize their own service opportunities for everybody to partake in and encourage more proactivity when doing good. To accomplish this, we hope to partner with local nonprofit organizations from RCC, UC Berkeley, and USC to provide service and donation opportunities within the app.

DoGood App Features

This is a code bundle for DoGood App Features. The original project is available at https://www.figma.com/design/6sOViGx6ABV7U0aTZF4G6b/DoGood-App-Features. The fully functional site is found here: https://calhacks25-gamma.vercel.app/

Project Structure

calhacks25/
├── src/               # React frontend application
├── api/               # Vercel serverless functions
├── voice-agent/       # LiveKit voice agent (Python)
└── ...

Setup

1. Install Dependencies

Frontend:

npm i

Voice Agent:

cd voice-agent
uv sync
cd ..

2. Claude API - ⚠️ Important: Requires Backend

The app uses Claude AI to:

Generate personalized community service activities
Verify task completion through photo analysis

🚨 CORS Limitation: Claude's API cannot be called directly from browsers due to security restrictions.

Option A: Use Mock Data (Recommended for Testing)

✅ Works out of the box - no setup needed!

All features functional
Perfect for development/demo
No backend required

Option B: Deploy with Backend (For Production)

To use real Claude API:

Deploy to Vercel (Recommended):

vercel deploy
vercel secrets add anthropic-api-key sk-ant-your-key-here

See CORS_SOLUTION.md for detailed setup instructions

Current Status: App uses mock data (works perfectly!). Deploy to enable live Claude AI.

3. LiveKit Voice Agent Setup

The "DoGood Companion" voice assistant requires LiveKit configuration:

Copy environment variables:
```
cp .env.example .env.local
```

Add your LiveKit credentials to .env.local:

LIVEKIT_API_KEY=your_api_key
LIVEKIT_API_SECRET=your_api_secret
LIVEKIT_URL=wss://your-project.livekit.cloud

Configure the voice agent:

cd voice-agent
# Update .env.local with same LiveKit credentials

See voice-agent/SETUP.md for detailed voice agent setup

Running the code

Quick Start (Recommended)

Run everything with a single command:

./start-dev.sh

This will start:

LiveKit voice agent (for DoGood Companion)
API server (for LiveKit token generation)
Vite development server (React frontend)

The frontend will be available at http://localhost:3000.

Manual Start (Alternative)

If you prefer to run services separately:

Terminal 1 - Voice Agent:

cd voice-agent
uv run agent.py dev

Terminal 2 - Frontend & API:

npm run dev

Features

Dynamic Activity Generation

Activities in the "Serve" section (currently uses mock data)
Can be personalized based on user preferences
Click "Refresh" to generate new activities
ℹ️ Requires backend for live Claude generation (see CORS_SOLUTION.md)

Photo Verification

Click the camera button to open your device's camera
Take a photo of your completed task
Verification (currently in demo mode)
Receive XP points for completed tasks
ℹ️ Requires backend for AI verification (see CORS_SOLUTION.md)

Voice Assistant (DoGood Companion)

Click "Speak with DoGood Companion" to start a voice conversation
Real-time audio interaction with AI assistant
Visual feedback showing listening/thinking/speaking states
Click "End Conversation" to close the assistant
Each button press starts a fresh conversation
ℹ️ Requires LiveKit setup and running voice agent (see voice-agent/SETUP.md)

Analysis

Compare with all teams

View

Metric

Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.

Technology

Found in codeClaimed only

CSSIn code
ExpressIn code
HTMLIn code
JavaScriptIn code
PythonIn code
ReactIn code
TypeScriptIn code
Next.jsClaimed
Tailwind CSSClaimed

7 of 9 appear in the indexed code. 2 claimed on Devpost could not be matched to code, which may simply mean the tool leaves no trace in the repository.

AI coding agents

Claude CodeCommits

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

853 KB

Source files

103

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.