DEV Community

Cover image for Airport: Manage Multiple AI Agent Terminals Happily ✈️
Vignaesh Ram A
Vignaesh Ram A

Posted on AI-assisted

Airport: Manage Multiple AI Agent Terminals Happily ✈️

AI coding agents have changed the way I develop software.

Instead of one assistant helping with one task at a time, I increasingly found myself running multiple agents in parallel — Claude Code, Codex, Devin, Antigravity, and others, each working on a different project or task.

And that's when I ran into a problem I wasn't expecting: managing the agents was becoming harder than using them.

The problem

My workflow quickly turned into something like this:

  • Open a terminal, start an agent
  • Switch workspace, open another terminal, start another agent
  • Switch back to check on the first one — is it still thinking, or has it been idle for 10 minutes?
  • Try to figure out which files it actually modified
  • Repeat, across 3-4 agents

VS Code terminals are great, but once you're running agents across multiple projects and workspaces, they all start looking identical. The question I kept asking myself was:

"Which files is this agent actually working on?"

The agent knows its workspace — but connecting agent → workspace → files → terminal meant constantly jumping between windows just to piece it together.

That's what led me to build Airport.

What Airport does

Airport is a VS Code extension for managing and monitoring multiple AI coding agents, side by side, in VS Code's own terminals.

A dedicated sidebar lists every agent terminal in one place, with a live status icon:

🔴 Needs you — blocked on a question or permission prompt
🟡 Working — thinking or running a tool
🟢 Done — finished its turn, idle

Instead of alt-tabbing to check state, you only switch to the terminals that actually need you.

Beyond status tracking, Airport also gives you:

  • Multi-workspace support in one window — run agents across different projects without juggling separate VS Code windows

  • A Files view that follows whichever terminal is selected — see exactly which folder an agent is working in, without touching your actual workspace folders

  • One-click agent launch — Claude Code, Codex, Devin, Antigravity, or a plain shell, each in a real VS Code terminal

  • Duplicate Terminal — spin up a copy of an existing terminal's setup instantly instead of manually recreating it
  • Duplicate Terminal with Another AI Agent — clone a terminal's workspace context but start a different agent in it, handy for comparing how Claude Code vs. Codex handles the same task

  • Notifications the moment a terminal needs your input

  • Session resume — reopen your workspace and pick up your previous terminals where you left off

How it works under the hood

Airport hooks into VS Code's shell integration API to watch terminal output and infer agent status, rather than parsing agent-specific output formats. That's also why status detection needs a shell with shell integration support (bash, zsh, fish, pwsh, cmd all work) — without it, the terminal still runs fine, it just won't show a status light.

It actually started life as a standalone Electron desktop app before I rebuilt it as a native VS Code extension. A separate window for terminal management stopped making sense once VS Code already had terminals I wanted to reuse directly.

Why "Airport"?

The name comes from the idea of an airport control tower controlling multiple terminals — multiple things happening at once, and you need a simple way to see what's happening, where, and what needs your attention right now.

✈️ Agents are taking off. Airport is the control tower.

Try it out

Free and open source:

Marketplace: https://marketplace.visualstudio.com/items?itemName=VignaeshRamA.airport
GitHub: https://github.com/vignaesh01/airport-vscode

I'd love to hear from developers already running multiple AI coding agents — how are you managing them today? Multiple terminals, separate VS Code windows, tmux, something else?

What's next

This is still early, and I'm figuring out where this workflow goes. As agents get more capable, I think the developer experience shifts from "I am coding with an AI assistant" to "I am orchestrating multiple AI agents." That means we need better tools for managing the agents themselves — not just better agents.

That's the problem I'm exploring with Airport.

#vscode #ai #aicoding #codingagents #claudecode #codex #developertools #opensource

Top comments (0)