What is Gemini CLI?
Gemini CLI is an open-source AI coding agent from Google that runs in your terminal as the gemini command. It reads your project, edits files and runs shell commands from plain-language instructions. You sign in with a Google account on first run, or use a Gemini API key.
Tallos is a desktop agent development environment that launches the real Gemini CLI inside isolated workspaces, next to your other agents, with terminals, diffs and a browser in one window.
How do I set up Gemini CLI in Tallos?
Go to Settings → Connections, install Gemini CLI from its card, sign in, and pick it as your default agent. An existing gemini on your PATH is detected automatically.
- 1
Install
Click Install. Tallos shows the official command before it runs:
brew install gemini-cliornpm install -g @google/gemini-clion macOS, andnpm install -g @google/gemini-clion Windows (Node.js required). - 2
Sign in
Click Sign in to start
geminiin a terminal and choose *Sign in with Google*. Or click Use API key: Tallos keeps aGEMINI_API_KEYin your computer's secure storage and passes it only to Gemini CLI when it starts it. - 3
Make it the default
In Settings → Agents, set Gemini CLI as the Default agent to have it pre-selected in every new workspace.
- 4
Start a task
Add a repository, click + for a new workspace, and describe the task. Tallos creates a git worktree on its own branch and launches Gemini CLI there with your prompt, in interactive mode.
How do I run multiple Gemini CLI sessions in parallel?
Create one workspace per task. Every workspace is a separate git worktree with its own branch and terminals, so each Gemini CLI session edits its own copy of the code.
- Side by side. Split panes to watch several sessions; the sidebar marks each one as working, waiting for you or done.
- Limits in view. The status bar reads Gemini CLI's local usage state and shows usage per limit window with time to reset, plus a warning at 80%.
- Review each branch. Open a workspace's diff, comment on lines and send the notes back to that session.
- Resume later. Past Gemini sessions show up in Agent Session History; Resume runs
gemini --resume <id>in the right folder.
Learn how the isolation works in parallel workspaces, and how to read agent output fast in diff review.
Run Gemini CLI in five workspaces at once.
How do I use Gemini CLI alongside Claude Code and Codex?
Put them in the same squad. A squad is a leader agent plus members; each slot picks its own agent, model and effort, and every member works in its own worktree.
- Best of three. Gemini CLI, Claude Code and Codex try the same objective separately; the leader compares the results and recommends one. See the best-of-three squad.
- Split by layer. With *Leader divides the work*, Gemini CLI can own the UI while other members take the API and the tests.
- Choose the model per member. Tallos applies the model you pick for a Gemini member when it launches it.
Squads run within the limits you set — 5 rounds and 4 members at once by default, up to 20 rounds and 12 members — and nothing merges without your approval. For a fair comparison of the three CLIs' approaches, read Claude Code vs Codex vs Gemini CLI.
Does Gemini CLI open in a chat view?
No. Gemini CLI opens in the terminal view in Tallos — its own interface, exactly as it looks in any terminal. The chat view is available for agents whose transcripts Tallos can read, such as Claude Code, Codex and Grok CLI.
Everything around the terminal still applies: isolated workspaces, splits, status in the sidebar, diff review and the built-in browser.
What does Tallos add to Gemini CLI?
| Capability | Gemini CLI on its own | Gemini CLI in Tallos |
|---|---|---|
| Parallel sessions | Open more terminals and set up git worktrees yourself | One workspace per task, each with its own worktree and branch |
| Install and sign-in | Run the install, then sign in on first run | Install, sign in or save an API key from Settings → Connections |
| API key handling | Export GEMINI_API_KEY in your shell | Kept in secure storage, passed only to Gemini CLI |
| Usage and limits | Checked inside the CLI | Status-bar readout with reset times and an 80% warning |
| Session status | Visible in its own terminal | Working / waiting / done across all sessions in the sidebar |
| Past sessions | gemini --resume from the right folder | Searchable history across agents, one-click Resume |
| Other agents | Separate tools | Same app; squads with a leader, members and best-of-N |
Frequently asked questions
Do I need a separate subscription to use Gemini CLI in Tallos?
No. Tallos uses the Google sign-in or Gemini API key you already use with Gemini CLI and does not resell model access.
How do I sign in to Gemini CLI from Tallos?
Click Sign in on the Gemini CLI card in Settings → Connections. Tallos opens gemini in a terminal, where you pick Sign in with Google; or save a GEMINI_API_KEY with Use API key.
Why does Tallos show Gemini CLI as 'sign-in unknown'?
Gemini CLI has no command that reports sign-in status, so Tallos checks for its cached Google credentials. If you use an API key or another method, the card can show 'Installed, sign-in unknown' even though the CLI works.
Can I run Gemini CLI and Claude Code on the same repo at the same time?
Yes. Each runs in its own worktree on its own branch, so they never edit the same files on disk. Compare their diffs and keep the one you want.
Does Gemini CLI work in Tallos on Windows?
Yes. On Windows 10+, the Connections page installs Gemini CLI with npm, so install Node.js first.
Can I pick the Gemini model for each squad member?
Yes. Each squad member has its own agent, model and effort setting, and Tallos applies the Gemini model you choose when it launches that member.
