The-Vibe-Company

Web & Mobile UI for Claude Code & Codex . Launch sessions, stream responses, approve tools. All from your browser / mobile

2,108
261
100% credibility
Found Feb 07, 2026 at 126 stars 17x -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
TypeScript
AI Summary

A dashboard and toolkit for creating, managing, and monitoring multiple AI coding agents that work on your project using your existing AI subscription.

How It Works

1
🔍 Find the coding helper

You discover this tool that lets you run and manage smart AI assistants to help improve your code project.

2
📦 Add it simply

You add the tool to your project workspace with one easy step.

3
🚀 Start your session

You launch the control center and choose your project folder to get everything ready.

4
Open the dashboard

A simple web page pops up showing your agents, messages, and controls in real time.

5
🤖 Create agents

You make one or more AI agents, giving each a name and role like coder or reviewer.

6
💬 Give tasks and approve

You assign jobs, chat with agents, watch them work, and approve their plans safely.

Code gets better

Your project now has improved code, tests, and fixes from the team of smart helpers.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 126 to 2,108 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is claude-code-controller?

Claude-code-controller is a TypeScript web UI that brings Claude Code out of the terminal via a reverse-engineered WebSocket protocol from the CLI. Run `bunx the-vibe-companion` to launch sessions in your browser at localhost:3456, stream token-by-token responses, and approve tools like Bash or file edits visually—no extra API keys needed, just your existing Claude subscription. Developers get multiple parallel sessions, hierarchical subagent views, and crash-resistant persistence.

Why is it gaining traction?

It solves Claude Code's biggest pains: no multi-session support, invisible tool calls, and lost context on crashes, all with claude web ui that feels like a native app. Streaming renders live, permissions range from auto-approve to manual blocks, and env profiles keep API keys project-specific without shell hacks. With 650 stars, it's hooking devs tired of terminal-only claude web agents and craving claude web access for quicker iterations.

Who should use this?

Backend engineers debugging with Claude Code who hate approving risky Bash tools blindly. Full-stack teams running parallel claude web api sessions on feature branches via integrated git worktrees. Solo devs prototyping web claude ai projects needing visual oversight without switching to github web ide or copilot chat.

Verdict

Grab it if you're deep in Claude Code—features deliver immediate wins for claude web github workflows, backed by solid tests and docs. At 1.0% credibility and 650 stars, it's early but promising; monitor for protocol changes as Anthropic evolves the CLI.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.