Home GitHub
1 / 10
Wee

Control Center for Claude Code

Wee

Run, Monitor & Control AI Agents

Live sessions. Real-time analytics. Multi-provider. One 15MB binary.

2026

The Question

What if managing AI agents
was effortless?

AI coding is exploding

Developers run dozens of Claude sessions daily. Without a control plane, you lose context, burn tokens, and have zero visibility into what your agents are doing.

The terminal isn't enough

CLI sessions disappear when you close the tab. No history, no metrics, no way to run agents side-by-side or share sessions across devices.

Wee is the missing layer

A self-hosted control center that wraps Claude Code with live sessions, real-time analytics, voice input, and multi-provider support — all in a single binary.

The Problem with AI Agents

"Running Claude Code in a terminal is powerful, but managing multiple sessions, tracking costs, and maintaining context across projects is a nightmare."
0 visibility into token spend
Lost context on every restart
1 session at a time in CLI
Fragmented terminal illustration

Meet Wee

A complete control plane for your AI coding workflow.

Wee

Live Sessions

WebSocket

Analytics

Real-time

Multi-Provider

6+ APIs

Permissions

YOLO mode

How Wee Works

Four pillars. One seamless experience.

1

Connect

Start a live agent session via browser. WebSocket streams every tool call in real-time.

2

Control

Granular permissions, YOLO mode toggle, provider selection — all from the UI.

3

Monitor

Real-time token usage, costs, tool executions, and session metrics on the dashboard.

4

Persist

SQLite storage keeps sessions, handoff context, and analytics across restarts.

Live Agent Sessions

1Prompt
2Execute
3Stream
4Render
Real-time
"Run multiple Claude sessions side-by-side in your browser. Every tool call — Bash, Read, Write, Edit — streams live via WebSocket."
ToolBash
ToolRead / Write
ToolEdit
ToolMCP Servers
Multi-provider illustration

Multi-Provider

Switch between Claude, DeepSeek, GLM, Kimi, Z.AI, or any Anthropic-compatible API — without changing a line of code.

Claude (Anthropic) DeepSeek GLM / ChatGLM Kimi (Moonshot) Z.AI Custom API

Real-Time Dashboard

Token usage. Costs. Tool calls. Git status. All live.

>_Wee
3Active
142Sessions
$4.82Today
SessionProviderTokensCost
refactor-apiClaude48.2K$1.24
fix-auth-bugDeepSeek124K$0.31
write-testsClaude67.8K$1.89

Built for Speed

A single 15MB Go binary. No Node.js, no Docker required. 50-100x faster than JavaScript alternatives.

GoSingle binary, zero dependencies
SQLitePersistent sessions & analytics
WebSocketReal-time streaming to browser
VoiceParakeet v3 speech-to-text (local)
DockerOptional — single command deploy
ARCHITECTURE 15MB

Single Binary Deployment

Install 
curl -fsSL https://raw.githubusercontent.com/
schlunsen/wee/
main/scripts/install.sh | bash
Run 
wee serve --port 3000
Docker 
docker compose up -d
Wee

Take control of your
AI coding workflow.

Live Sessions / Multi-Provider / Real-Time Analytics
View on GitHub
curl -fsSL https://raw.githubusercontent.com/schlunsen/wee/main/scripts/install.sh | bash