VibeHandler › Guides
How to run multiple Claude Code agents on Windows
Running several Claude Code agents at once is the fastest way to get more done — one agent builds the API while another writes tests and a third refactors the frontend. Here's how to do it on Windows, and how to stop losing sessions when your machine restarts.
The quick answer
To run multiple Claude Code sessions in parallel on Windows, open one terminal per task and start claude in each project folder — or use a session manager like VibeHandler that puts every agent in one window, shows live status and cost, and restores every session automatically after a crash or reboot. VibeHandler is free for up to 3 agents.
Option 1 — Multiple terminals (manual)
The zero-install way:
- Install Node.js and the Claude Code CLI, then sign in with your Anthropic Claude subscription.
- Open a new terminal (Windows Terminal, PowerShell or your IDE terminal) for each task.
- In each terminal,
cdinto the project folder and runclaude. Each terminal is now an independent agent. - Switch between them with tabs or split panes.
The catch: juggling many terminal windows gets chaotic fast, you can't see at a glance which agent is waiting for input, and if Windows restarts (or a window closes) you lose track of which session was where — resuming means hunting for the right claude -r <session-id>.
Option 2 — A session manager (VibeHandler)
A session manager solves the chaos and the data-loss problem:
- Download and install VibeHandler for Windows 10/11.
- Open one pane per task. Each pane is a real Claude Code terminal — no wrapper — running side by side in a grid, with live status (working / waiting / done), cost and git branch.
- Get a phone notification the moment any agent needs your input.
- Optionally tell the conductor agent what you want; it starts, tasks and monitors the other agents for you.
Crucially, every session is cataloged and backed up automatically. After a crash, power cut or Windows update, VibeHandler restores all panes and resumes each conversation with full context using claude -r — so you never lose a session.
Manual terminals vs. VibeHandler
| Multiple terminals | VibeHandler | |
|---|---|---|
| Agents in one window | ✕ separate windows | ✓ grid |
| See who's waiting | ✕ check each one | ✓ status + phone push |
| Survives a reboot | ✕ resume by hand | ✓ auto-restore |
| One agent orchestrates the rest | ✕ | ✓ conductor |
| Search across old sessions | ✕ | ✓ fleet memory |
| Cost | Free | Free ≤3 · Pro €5/mo |
How many agents can you run at once?
There's no hard limit in Claude Code itself — it's bounded by your machine's memory and your own attention. Most people comfortably run 3–6 agents in parallel. VibeHandler's free tier covers up to 3 concurrent agents; Pro removes the limit and adds the conductor so you can coordinate more without watching every one.
Does running many agents cost extra?
No extra API or token cost beyond what the agents themselves use — VibeHandler drives the Claude Code CLI on your existing Anthropic subscription. There's no separate API key. Each agent's live token use and cost is shown per pane so you always know what's running.
FAQ
Can I recover a Claude Code session after a crash on Windows? Yes. With VibeHandler every session is auto-cataloged and backed up; after a restart all panes are restored and resumed with full context. Manually, you can also run claude -r with the session ID if you still have it.
Is this a Conductor alternative for Windows? Yes — Conductor is macOS-only, and VibeHandler brings the same "many agents, one orchestrator" idea to Windows, plus crash-safe recovery.
Do I need a Claude subscription? Yes. VibeHandler is a front-end for the Claude Code CLI, which runs on your own Anthropic Claude subscription.
Download VibeHandler for Windows
Free for up to 3 agents · Windows 10/11 · runs on your Claude subscription.