termdeck
Every Claude Code, Codex, and Grok session. One console.
Termdeck is a free coding agent web UI for Claude Code, Codex CLI, and Grok CLI. Three engines in one window, running on your own machines. Sessions from every machine land in the sidebar, the ones you started in a terminal included, with no tmux panes and no SSH tabs. And when a Claude account hits its rate limit, switch to another in one click; the same chat keeps going.
14 days of Pro, free · then the free plan · no card required
A Termdeck run, drawn: a prompt goes to a coding agent CLI on your own
machine, every turn is appended to that CLI's own session file on disk, a tool call
that needs permission waits for you to answer it in the browser or on your phone, and
the whole thing reaches you over the machine agent's outbound connection, with your
other machines still running their own sessions alongside it. Because the transcript is
the CLI's own file, claude --resume and codex resume reopen the
same session in your terminal afterwards.
A Claude Code web UI that runs on your own machines
Connect a machineTerminals
4 terminals, 4 models, 4 tasks. All at the same time.
Tile any four sessions onto one canvas and watch them all work. Different engines, different machines, different repositories, each on its own turn: you read the one that is interesting and the other three keep going. Alt+1–9 moves between them; a tool call that needs approval raises its own window.
Before / after
Terminal babysitting vs a calm agent fleet.
In the terminal
- Tabs, panes, and lost scrollback
- Approvals stuck until you notice the prompt
- No fleet view across machines
- A rate limit hits and you wait it out
With Termdeck
- One coding agent session manager for every local agent
- Push + phone when a tool needs you
- Laptop, desktop, VM: same dashboard
- Switch accounts in one click, same chat
A control plane, not another sandbox
Your agents already know your machine. Now you can reach them.
Termdeck sits on top of the real Claude Code, Codex, and Grok CLIs. Your repository, shell, environment, MCP servers, subscriptions, and native session files stay where they are.
The coding agent dashboard
Three engines, every machine, one window.
Run Claude Code, Codex, and Grok in parallel.
Design the UI with Codex while Claude writes the backend. Send the easy tasks to one engine and the hard ones to another. One project on the laptop, another on the work PC, all in the same window.
Approve Claude Code, Codex, and Grok from your phone.
Install the PWA, get a push notification when a tool needs approval, and unblock the agent before your coffee gets cold.
Review the work, not the scrollback.
Open changed files, scan additions and removals, and understand what an agent touched before you take over.
Love the terminal? Keep it.
The Terminals view renders every chat as a pane in one tmux-style grid: every machine, every engine, one screen. All the terminal energy, none of the tmux setup.
Allow just the commands that matter.
Auto-allow the routine, approve the risky from wherever you are, and set a hard spend ceiling per run. The agent moves fast; nothing moves outside your rules.
One-click account switching
Two Claude subscriptions? Stop waiting for limits.
Connect every Claude, OpenAI, and xAI account you have. Termdeck shows each account's live rate limits side by side, and when one runs dry mid-session, you switch in one click and keep typing. No logout, no lost context, no countdown clock.
Live limit meters per account, and switching keeps the same chat going
Right model, right task
Save up to $500 a year by routing work to the right model.
Not every task deserves your most expensive plan. Keep the premium subscription for the hard problems, send the routine work to a cheaper engine, and let the usage dashboard prove the split is working.
Published prices, August 2026: Claude Max $100–$200/mo, ChatGPT Plus $20/mo. Your saving depends on your mix. Model usage bills to Anthropic, OpenAI, or xAI, never to Termdeck.
Watch it work
The whole control plane, in under a minute.
A blocked run finds you on your phone and gets approved. Then the fleet board, three engines in parallel, the diff, what it cost, and the terminal resume that still works.
How the local web UI works
Your CLI does the work. Termdeck makes it visible.
No new model account and no proprietary agent runtime. Connect the machines you already use and pick up from any screen.
Add a machine.
Sign in, create a one-time machine token, and run the line the dashboard hands you: curl -fsSL https://termdeck.io/install.sh | TERMDECK_AGENT_TOKEN=agt_… sh. On Windows, the same line via install.ps1.
Use your existing agent auth.
Claude Code, Codex, and Grok run with the subscriptions and MCP config already on each machine.
Open any browser.
Start a session, stream its work, approve tools, send screenshots, steer a run mid-turn, and review changes. Sessions stay on disk.
Every token, every model, every account, one ledger
Track Claude Code and Codex usage by project.
Termdeck rolls Claude Code, Codex, and Grok token usage into project totals, estimates API-equivalent cost, and lets you drill down to the exact run.
And the rest of the console
It keeps going.
Filter every chat on every machine by prompt, project, or host, or press Ctrl+R in a chat and pull back any prompt you've sent it. Dictate instead of typing. Paste a screenshot at the bug. Claude turns that edit files leave checkpoints: restore the files in a click, or branch the chat from that point. Cap what a single turn may spend. Open the MCP panel and see every server a chat runs with: status, errors, reconnect. Edit the project's CLAUDE.md or AGENTS.md from a chat's menu, or type # in the composer to append a note. Type the next instruction mid-turn and it queues, or steers the run, on engines that steer. A dev server the agent left running gets its own pane: live output, elapsed time, one Stop. Install the console as an app, and approvals ride the push notification itself. And when the process driving a turn dies without a word, the console notices the silence and hands the chat back.
Choose the route that matches your search
A Claude Code UI, a Codex web app, or a Grok CLI dashboard? Start here.
Termdeck is the shared browser layer for all three local CLIs. These focused guides explain how it compares with the terminal, the official cloud products, and multi-pane agent workflows.
Claude Code web UI, on your own machine.
Run the real Claude Code CLI from a browser or phone while keeping local context, auth, and terminal resume.
OpenAI CodexA visual dashboard for Codex CLI rollouts.
Read Codex transcripts and diffs in the web, then manage them beside every Claude Code and Grok session.
xAI GrokGrok CLI, driven from the browser.
Start a Grok Build session, stream work, approve tools, and switch model or effort.
Agent managerA calmer alternative to tmux for AI agents.
Replace pane hunting with session status, approvals, usage, search, and a fleet view built for coding agents.
Every account starts on Pro. No card.
Sign up and you get 14 days of Pro, free: your whole fleet, unlimited sessions at once. Then you drop to Free automatically. No card is asked for at any point, and nothing is deleted when it ends. Bring your own Claude, OpenAI Codex, and xAI (SuperGrok) subscriptions. Termdeck is the control plane.
- Unlimited approvals, stops, and transcript reading
- Push notifications when an agent needs you
- 30 browser turns a month
- 1 connected machine, 1 session at a time
- Claude Code + Codex + Grok, full history
- Auto-approve rules, so you stop answering the same prompt
- Queue follow-up turns and walk away
- Unlimited browser turns
- Checkpoints, spend analytics, MCP, voice, search
- 2 connected machines, unlimited sessions at once
- Everything in Starter, on every machine you own
- 10 connected machines, unlimited sessions at once
- Fleet board and search across the whole fleet
- Early access to new engines and features
Cancel any time; a cancelled plan drops back to Free rather than locking you out. Have a creator code? Redeem it on your account page. Model tokens and API usage bill to Anthropic, OpenAI, or xAI, not to Termdeck. Need more than 10 machines? Email [email protected] for custom fleet pricing.
The short answers.
Everything about running local Claude Code, Codex, and Grok from the browser. Can't find it? Reach out.
Contact usYes. Termdeck is a browser UI for the Claude Code CLI running on your machine. It streams the local transcript and adds tool approvals, diffs, project usage, session search, and phone access without moving the agent into a separate runtime. Read the Claude Code web UI guide →
Termdeck provides one by reading Codex CLI rollout files from disk and rendering the transcript, tool calls, and changed files in the browser, beside your Claude Code sessions. Read the Codex web UI guide →
Yes. Termdeck runs xAI's Grok CLI as a third engine over the Agent Client Protocol. Start a Grok Build session, stream the transcript, approve tool calls, and switch model or reasoning effort mid-run. Grok sessions use grok's own session files, so a terminal and the browser can share one session. Read the Grok CLI guide →
Claude Code on the web runs tasks on Anthropic-managed cloud infrastructure. Termdeck drives Claude Code on machines you own, so it can use your local repository, services, environment variables, credentials, and MCP servers. Compare Claude Code on the web vs Termdeck →
Yes. Connect multiple accounts per engine, see each account's live rate limits side by side, and switch in one click when one hits a limit. The session keeps going under the account you switched to.
Yes. Install Termdeck as a PWA to follow Claude Code, Codex, and Grok sessions, send prompts and screenshots, and approve or deny tool calls. Push notifications surface waiting approvals when you are away from the computer.
The agents and their native session data stay on your machines. Termdeck starts loopback-only by default, and its remote connection options are designed to fail closed unless you configure authentication deliberately.
Yes. Termdeck reads and writes the same session files as the CLIs, so you can continue with claude --resume or codex resume. Terminal-started sessions also appear in the browser.
Yes. Connect macOS, Linux, or Windows machines and browse their active and historical sessions from one fleet board. Useful for a laptop, desktop, VM, home server, or remote development box.
No. It is the UI and control layer for the official CLIs: Claude Code, Codex, and Grok. You keep your existing Anthropic, OpenAI, and xAI access, local tools, configuration, and model choices.
Put your coding agents one tab away.
Open the web dashboard for Claude Code, Codex CLI, and Grok CLI, connect your machines, and pick up any session from any screen: desk, laptop, or phone.