EVOX - Mission Control

A real-time dashboard for orchestrating AI agents that collaborate on software development tasks. Agents pick up tickets, write code, commit, and report back. You review and steer.

What is this?

EVOX is a mission control interface for managing a team of AI coding agents. Instead of one AI chatbot doing everything, EVOX splits work across specialized agents that coordinate autonomously:

• Max (PM) - Creates tickets, tracks progress, writes retros
• Sam (Backend) - Convex functions, API logic, data layer
• Leo (Frontend) - React components, UI, styling

Linear is the source of truth. EVOX syncs every 2 minutes, tracks status changes, and surfaces everything in a live dashboard. The human (you) jumps in when agents need direction, architecture decisions, or feedback.

Features

• Dashboard - Task counts, agent status, recent activity feed
• Tasks Kanban - Backlog / Todo / In Progress / Done columns, synced from Linear
• Daily Standup - Per-agent breakdown of completed, in progress, and blocked work
• Activity Feed - Timestamped log of every status change
• Linear Sync - Bi-directional sync on a 2-minute cron, with automatic activity creation on status changes
• Agent Cards - Real-time status, role, and current task for each agent

Stack

Layer	Tech
Frontend	Next.js 15, React, shadcn/ui, Tailwind CSS
Backend	Convex (realtime database + serverless functions)
Project Management	Linear (via API sync)
Deploy	Vercel
Theme	Dark mode

Architecture

Linear API
    │
    ▼ (every 2 min)
┌─────────────────────────────┐
│  Convex Backend              │
│  ├── linearSync:syncAll      │
│  ├── tasks table (39)        │
│  ├── activities table (43)   │
│  ├── agents table (3)        │
│  └── projects table (4)      │
└─────────────┬───────────────┘
              │ useQuery (realtime)
              ▼
┌─────────────────────────────┐
│  Next.js 15 Frontend         │
│  ├── Dashboard               │
│  ├── Tasks Kanban            │
│  ├── Daily Standup           │
│  ├── Activity Feed           │
│  └── Agent Cards             │
└─────────────┬───────────────┘
              │
              ▼
         Vercel (prod)

Getting Started

Prerequisites

• Node.js 18+
• Convex account
• Linear account + API key

Setup

# Clone
git clone https://github.com/sonpiaz/evox.git
cd evox

# Install
npm install

# Set up Convex
npx convex dev

# Environment variables
cp .env.example .env.local
# Add your NEXT_PUBLIC_CONVEX_URL and LINEAR_API_KEY

Environment Variables

Variable	Where	Description
NEXT_PUBLIC_CONVEX_URL	.env.local + Vercel	Convex deployment URL
LINEAR_API_KEY	Convex env	Linear API key for sync

# Set Linear key in Convex
npx convex env set LINEAR_API_KEY your_key_here

Run

# Terminal 1: Convex backend
npx convex dev

# Terminal 2: Next.js frontend
npm run dev

Open http://localhost:3000.

Project Status

MVP: 83% complete (30/39 issues done)

Phase	Status
P1 Foundation (schema, CRUD, UI shell)	Done
P2 Communication (messages, notifications)	Done
P3 Heartbeat/CLI	Done
P4 Linear Sync	Done (4 polish bugs remaining)
P3 Execution Engine	Backlog
P5 Advanced (Slack, GitHub auto-commit)	Partial

How It Was Built

This project was built using AI agents orchestrated through Claude and Cursor:

• Claude (claude.ai) as Max the PM - managing Linear tickets, writing specs, diagnosing bugs from screenshots, updating project state
• Claude Code / Cursor as Sam and Leo - writing and committing code based on specs from Max
• Human (Son) as technical lead - reviewing output, providing feedback, making architecture decisions

The feedback loop: Linear ticket created, agent picks it up, codes the solution, commits, deploys. Human reviews screenshots, files bugs if needed, agents fix. Rinse and repeat.

What's Next

• Execution Engine - Agents autonomously pick tasks, run code, commit to GitHub
• Auto-PR - Agents create pull requests, human approves
• Slack Integration - Notifications on task completion and blockers

License

MIT

Author

Son Piaz (@sonpiaz)

Built with AI agents, for AI agents.