Run Claude Code in parallel — Tallos

What is Claude Code?

Claude Code is an agentic coding tool from Anthropic that runs as a command-line program called claude. You describe a task in plain language; it reads your repository, edits files and runs commands in your terminal. It signs in with a Claude account or uses an Anthropic API key.

On its own, Claude Code is one session in one terminal. Tallos is a desktop agent development environment that runs Claude Code as it is — the same CLI, the same ~/.claude folder, the same CLAUDE.md files — and adds what you need to run many sessions at once and review what they produce.

How do I set up Claude Code in Tallos?

Open Settings → Connections, install Claude Code from its card, sign in, then make it your default agent. If claude is already on your PATH, Tallos detects it and you can skip the install.

1. 1

   Install

   Click Install on the Claude Code card. Tallos shows the vendor's official command before running it: brew install --cask claude-code or npm install -g @anthropic-ai/claude-code on macOS, and npm install -g @anthropic-ai/claude-code on Windows (Node.js required).

2. 2

   Sign in

   Click Sign in to run claude auth login in a terminal, or Use API key to save an ANTHROPIC_API_KEY in your computer's secure storage. Tallos gives the key only to Claude Code, each time it starts it, and never shows it again.

3. 3

   Make it the default

   In Settings → Agents, set Claude Code as the Default agent. It is pre-selected every time you open a new workspace.

4. 4

   Start a task

   Add a repository, click + to create a workspace and describe the task. Tallos creates a git worktree on its own branch and launches claude with that folder as its working directory.

Setting up more than one agent? Follow connect your agents for every CLI on the Connections page.

How do I run multiple Claude Code sessions at the same time?

Create one workspace per task. Each workspace is a separate git worktree with its own branch, files and terminals, so several Claude Code sessions can change the same repository without overwriting each other.

• No shared working directory. Each session runs inside its own worktree, so installs, builds and tests stay with that task.
• Watch them side by side. Split panes in any direction. The sidebar shows which sessions are working, waiting for you or done.
• Subagents stay visible. Background subagents and teammates appear as child rows under the lead Claude Code session.
• Review per branch. Every workspace has its own diff. Comment on a line and send the feedback back to the session that wrote it.
• Resume later. Agent Session History lists past Claude Code sessions; Resume runs claude --resume <id> in the right folder.

The full walkthrough is in how to run multiple Claude Code sessions. For how the isolation works, see parallel workspaces.

Run your first parallel Claude Code sessions today.

Get Tallos →

Can I use more than one Claude account and watch my limits?

Yes. Tallos manages several Claude accounts under Settings → AI Provider Accounts and switches the active one in one click, with no re-login. New sessions use the selected account; sessions already running keep theirs until you restart them.

The status bar reads Claude Code's usage state and shows how close you are to each limit window, with the time to reset. A warning appears when you cross 80% of a limit, so you see it before a session stalls. The readout is only as fresh as the data the CLI records.

How do I use Claude Code alongside Codex, Gemini CLI and other agents?

Run them side by side in the same project, or put them in a squad. A squad is a leader agent plus members, and each slot picks its own agent, model and effort level.

• Claude Code as leader. It splits the objective, hands out tasks, checks results and loops within the limits you set — 5 rounds and 4 members at once by default, up to 20 rounds and 12 members.
• Claude Code as reviewer. In the *Build + review* template, another agent builds and Claude Code reviews until it approves.
• Best of N. In *3 attempts, pick the best*, Claude Code, Codex and Gemini CLI try the same objective in separate worktrees, and the leader recommends one.

Every member works in its own worktree. Nothing merges until you Accept; Discard keeps or deletes the worktrees. For a side-by-side of the two most common CLIs, read Claude Code vs Codex.

Does Claude Code open in a chat view or a terminal?

In a chat view, by default. Tallos reads Claude Code's transcript and lays out messages, tool calls and questions for reading, while the real CLI keeps running underneath as the source of truth.

• View terminal and Back to chat in the pane header switch between the chat and the raw terminal UI at any time.
• The composer supports / for slash commands and skills, file and image attachments, and a model pill whose choices come from the Claude CLI installed on that machine.
• When Claude Code asks a structured question, the chat shows it as a card you answer and Submit.
• Prefer the terminal? Set Settings → Agents → Agent View to Terminal.

More detail on the composer and transcript in native chat.

What does Tallos add to Claude Code?

Tallos does not replace or modify Claude Code. It adds the environment around it: workspaces, views, accounts and review.