A flexible terminal grid for multi-agent AI workflows

Architect

https://github.com/user-attachments/assets/a4e28a63-557a-44f3-9bae-47b2fd0a5dc6

A terminal built for multi-agent AI coding workflows. Run Claude Code, Codex, or Gemini in parallel and see at a glance which agents need your attention. See more in my article.

Built on ghostty-vt for terminal emulation and SDL3 for rendering.

Why Architect?

Running multiple AI coding agents is the new normal. But existing terminals weren't built for this:

• Agents sit idle waiting for approval while you're focused elsewhere
• Context switching between tmux panes or tabs kills your flow
• No visibility into which agent needs attention right now

Architect solves this with a grid view that keeps all your agents visible, with status-aware highlighting that shows you instantly when an agent is awaiting approval or has completed its task.

[!WARNING] This project is in the early stages of development. Use at your own risk.

The application is experimental and may have bugs, stability issues, or unexpected behavior.

Features

Agent-Focused

• Status highlights — agents glow when awaiting approval or done, so you never miss a prompt
• Agent session persistence — when you quit Architect, any running Gemini, Codex, or Gemini agents are gracefully terminated and their session IDs saved; on next launch the agents resume automatically where they left off
• Dynamic grid — starts with a single terminal in full view; press ⌘N to add a terminal after the