AI agents

Several agents, several terminals, no overview

Project tabs, agents in terminals, tasks and crons, agent health in real time.

A run through the dashboard, demo data: the project tabs, one tab’s agents, tasks and crons, an agent started with one click and its health updating, every project’s crons, and the agents’ log.
Dashboard, demo data: the counters (active tabs, running agents, pending tasks, alerts), one card per project tab and the health of every agent.
Agents page, demo data: each agent with its model, its status and its last heartbeat; an agent just started from its Start button, its card now shows Stop.
A project tab, demo data: its agents, its recent tasks with type, status and priority, and its crons.
Crons page, demo data: every project’s scheduled jobs with their cron expression, their project and their run count, active or paused.

The problem

Running several AI coding agents across projects means juggling terminals with no view of their health or tasks.

What I built

A Next.js dashboard on a local Supabase: agents run in tmux sessions, their health streams in real time, and their tasks and crons are listed per project.

Technical details

Generic typed hook over Postgres changes, with dynamic filters
28 tables with updated_at triggers and indexes

Stack

Next.js 16
React 19
TypeScript strict
Supabase Realtime
tmux

The result

An internal prototype with a versioned 28-table Postgres schema and a typed real-time hook.

A similar project in mind?

Tell me about your business and when you need it. You get a proposal by email.

I want this system

More projects

ERP & CRM · Shared WhatsApp deskThree salespeople on one number, two prices for the same customerView the project