Run your coding agent from your phone.

Coding agents need input every 10-60 minutes. If you're not at your desk, they just sit there. Sled lets you run them from your phone, with voice so you can go faster.

Works with the tools you already use
Claude Code Claude Code
OpenAI Codex
Gemini CLI Gemini CLI
Why voice?

Talking is faster than typing. You also need to know when your agent is done. Terminals can't do two-way voice, so Sled runs in the browser.

Your computer stays home.
You don't have to.

1

You talk

Sled transcribes and sends it to your agent over Tailscale.

2

Agent works

Runs locally on your computer. Code never leaves your machine.

3

You hear back

Response converted to speech. You hear what it did.

Connected via Tailscale
"Add dark mode to the settings page"
I've added a toggle in SettingsPage.tsx and created a ThemeContext. Want me to add the CSS variables too?

What you get

🔔

Notifications

Your agent finishes a task or hits a wall. You get a ping. It tells you what happened.

🎙️

Voice input

Talk instead of type. Handles camelCase and function names.

🗣️

Voice output

Responses read aloud. 300+ voices to choose from.

🛏️

Works away from your desk

Couch, bed, walking around. Your laptop stays where it is.

🎧

Hands-free

AirPods in, phone in pocket. No screen needed.

🏠

Code stays local

Your agent runs on your machine. Code never leaves.

🌐

Tailscale

Encrypted connection. No port forwarding.

Open source

Fork it, modify it, self-host it.

Setup takes 5 minutes

You need Tailscale and a coding agent (Claude Code, Codex, or Gemini CLI) on your computer.

# Clone the repo git clone https://github.com/layercodedev/voice-code.git
# Run the setup script cd voice-code && ./setup.sh
View on GitHub
🌳 ☀️ 🚶 💻 🌿

Touch grass while you code

Actual footage of a Sled user touching grass