Project Info

CappyCoding

Inspiration

We wanted to build upon the tried and true rubber ducky method of debugging, where a developer explains their thought process to a "rubber ducky", forcing themselves to walk through their system, start to finish. We wanted to create a tool that can be used for this as well as some flair!

What it does

We provide a cute screen, ideally held up by our mascot, that shows developers their coding stats such as GitHub Pull Requests, GitHub Actions, commits or more specific to vibe coding like Claude token usage etc.

How we built it

We connected an ESP32 board to an E-Ink display, and used Rust to program the TUI widgets. A Go server then complements the embedded device, which polls for live updates and stats. The device runs standalone due to using the Wi-Fi/BLE stack of the esp32.

Challenges we ran into

Since some of the teammates were new to Rust, some time was spent on learning before building. It was also our first time trying to understand the Wi-Fi and Bluetooth protocols at a deeper level, and then trying to find examples which we could build on for those technologies was very difficult. There was a lot of trail and error.

Accomplishments we're proud of

We are extremely proud of the fact that we built a standalone embedded device that functions on ubiquitous protocols. Just seeing it contact our backend while only connected to power, and then pull that data down and then displaying it; it was incredibly rewarding.

What we learned

We learned so much about interfacing with networking protocols as well as more experience with embedded software development!

What's next

We would love to spend more time on the project and add more features, and even design a nice case so it could be a practical desk-gadget!

CappyCoding

A Tauri-based desktop application with voice-enabled AI assistant powered by LiveKit and Claude.

Features

🎙️ Voice AI Assistant: Talk to Claude using natural speech
🔊 LiveKit Integration: Professional-grade WebRTC for real-time communication
🤖 Claude Sonnet 4.5: Advanced AI responses via Anthropic
📊 Usage Metrics: Track Claude API usage and costs
🎨 Modern UI: Built with Svelte 5 and Tauri 2.0
🔧 Easy Configuration: UI-based setup for API keys and agent management

Architecture

┌─────────────────────────────────────────────────────────────┐
│                      Tauri Desktop App                      │
│                     (Svelte 5 Frontend)                     │
├─────────────────────────────────────────────────────────────┤
│  🎤 Voice Input  →  LiveKit  →  Python Agent  →  🔊 Voice   │
│                                      ↓                       │
│                            LiveKit Inference                 │
│                       (Deepgram STT + Cartesia TTS)         │
│                                      ↓                       │
│                            Claude AI (Anthropic)             │
└─────────────────────────────────────────────────────────────┘

Quick Start

Prerequisites

Node.js 18+ and bun (for frontend)
Rust (for Tauri)
Python 3.13+ (for voice agent)

1. Get API Keys

You'll need:

LiveKit Cloud account: https://cloud.livekit.io/ (free tier available)
Anthropic API key: https://console.anthropic.com/

2. Launch the App

cd capycoding-app
bun install
bun run tauri dev

3. Configure Voice Agent (in the app)

Find the "🤖 Voice Agent Configuration" panel
Enter your LiveKit credentials (URL, API Key, API Secret)
Enter your Anthropic API key
Click "💾 Save Configuration"
Click "▶️ Start Agent"

4. Start Talking!

Go to "🎙️ Connect to Voice Session"
Enter a participant identity and room name
Click "Connect to Voice Session"
Start speaking naturally!

Project Structure

CappyCoding/
├── agent.py              # Python voice agent (LiveKit + Claude)
├── env/                  # Python virtual environment
├── capycoding-app/       # Tauri desktop application
│   ├── src/              # Svelte frontend
│   └── src-tauri/        # Rust backend
├── capycoding-esp/       # ESP32 firmware (optional)
├── ble-types/            # Bluetooth type definitions
├── server/               # Go metrics server
└── docs/
    ├── FRONTEND-CONFIG.md  # UI configuration guide
    ├── README-AGENT.md     # Agent technical details
    └── QUICKSTART.md       # Quick start guide

Components

Voice Agent (`agent.py`)

Python agent using LiveKit's agent framework:

STT: Deepgram (via LiveKit Inference)
LLM: Claude Sonnet 4.5 (via Anthropic plugin)
TTS: Cartesia (via LiveKit Inference)
VAD: Silero voice activity detection

Configuration stored in:

~/.config/capycoding/agent_config.json (UI-saved)
.env file (manual configuration)
Environment variables (highest priority)

Tauri App (`capycoding-app/`)

Desktop application featuring:

Voice agent configuration and management
LiveKit voice session connection
Claude usage metrics tracking
Real-time audio streaming

Built with:

Frontend: Svelte 5, Vite, LiveKit Client SDK
Backend: Rust, Tauri 2.0, taurpc for IPC

Documentation

QUICKSTART.md - Get started in 1 minute
FRONTEND-CONFIG.md - UI configuration guide
README-AGENT.md - Agent technical details

Costs

Using LiveKit Inference simplifies billing to just 2 providers:

LiveKit Cloud (includes STT + TTS):

Deepgram STT: $0.0043/minute
Cartesia TTS: $0.045/minute
LiveKit transfer: Free tier includes 50 GB/month

Anthropic:

Claude Sonnet 4.5: ~$3/$15 per million tokens

Typical usage: ~$0.26/hour of conversation

Development

Running the Agent Manually

cd /path/to/CappyCoding
source env/bin/activate
python agent.py dev

Building the Tauri App

cd capycoding-app
bun run tauri build

Installing Agent Dependencies

Already installed in /env, but to reinstall:

cd /path/to/CappyCoding
python -m venv env
source env/bin/activate
pip install livekit "livekit-agents[anthropic,silero,deepgram,cartesia]"

Environment Variables

The agent loads configuration in this priority:

Environment variables (highest)
UI-saved config (~/.config/capycoding/agent_config.json)
.env file (lowest)

Required variables:

LIVEKIT_URL=wss://your-project.livekit.cloud
LIVEKIT_API_KEY=APIxxxxxxxxxx
LIVEKIT_API_SECRET=xxxxxxxxxxxxxxxx
ANTHROPIC_API_KEY=sk-ant-xxxxxxxxxx

Troubleshooting

Agent won't start

Verify API keys are correct (no extra spaces)
Check that /env virtual environment exists
Try running manually to see error messages

Can't connect to voice session

Ensure agent is running (check status in UI)
Verify LiveKit credentials in agent configuration
Check that participant identity and room name are filled in

No audio response

Check LiveKit Inference is enabled (default on new projects)
Verify microphone permissions in your OS
Check browser/Tauri audio settings

See QUICKSTART.md for more troubleshooting tips.

License

[Your License Here]

Contributing

[Your Contributing Guidelines Here]

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
GoIn code
HTMLIn code
JavaScriptIn code
PythonIn code
RustIn code
SvelteIn code
Tailwind CSSIn code
TypeScriptIn code

9 of 9 appear in the indexed code.

AI coding agents

No AI coding agent signals were found in this repository.

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

Codebase size

Source size

584 KB

Source files

412

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

Repository

Sluggish-Solutions/CappyCoding

464 files · 1.4 MB · @ b8aa5e8

Structure

Interface
360 files · 78%
Screens, components and styles rendered to the user.
API & routing
16 files · 3%
Request entry points: routes, handlers and controllers.
Application logic
37 files · 8%
Domain rules, services and shared utilities.
+1 more

Supporting

Layers are inferred from where files sit in the tree, not from reading the code. A project that names its directories unconventionally will read oddly here — open the file browser to check anything the diagram implies.

Languages

Svelte60%
Rust10%
Markdown8%
Go8%
TypeScript8%
Python3%
Other (5)3%

Share of indexed source by file size. Binary and vendored files are excluded.

Dependencies

capycoding-app/package.json

npm · 41

@livekit/components-core
@livekit/components-styles
@lucide/svelte
@tauri-apps/api
@tauri-apps/plugin-opener
@thisux/sveltednd
@xenova/transformers
clsx
livekit-client
neverthrow
runed
tailwind-merge
tailwind-variants
taurpc
tw-animate-css
+26 more

capycoding-esp/Cargo.toml

cargo · 34

ble-types
bt-hci
critical-section
display-interface-spi
embassy-executor
embassy-futures
embassy-net
embassy-sync
embassy-time
embedded-graphics
embedded-hal-bus
embedded-io
embedded-io-async
embedded-storage
esp-alloc
esp-backtrace
esp-bootloader-esp-idf
esp-hal
+16 more

server/go.mod

go · 20

github.com/google/go-github/v55
github.com/labstack/echo/v4
golang.org/x/oauth2
+17 more

capycoding-app/src-tauri/Cargo.toml

cargo · 16

anyhow
ble-types
btleplug
chrono
dirs
jsonwebtoken
log
reqwest
serde
serde_json
specta
tauri
tauri-plugin-opener
taurpc
tokio
+1 more

ble-types/Cargo.toml

cargo · 4

heapless
postcard
serde
uuid

Declared in the repository’s manifests at the indexed commit. A declared package is not proof it is used, and runtime dependencies are listed first.

CappyCoding

Inspiration

What it does

How we built it

Challenges we ran into

Accomplishments we're proud of

What we learned

What's next

CappyCoding

Features

Architecture

Quick Start

Prerequisites

1. Get API Keys

2. Launch the App

3. Configure Voice Agent (in the app)

4. Start Talking!

Project Structure

Components

Voice Agent (agent.py)

Tauri App (capycoding-app/)

Documentation

Costs

Development

Running the Agent Manually

Building the Tauri App

Installing Agent Dependencies

Environment Variables

Troubleshooting

Agent won't start

Can't connect to voice session

No audio response

License

Contributing

Technology

AI coding agents

Codebase size

Structure

Interface

API & routing

Application logic

Languages

Dependencies

Voice Agent (`agent.py`)

Tauri App (`capycoding-app/`)