termdeck
Compare
Each page says what the other tool does well, where it stops, and what Termdeck adds. Your agents run on your own machines either way; the difference is how you reach them, how many machines you see at once, and where approvals land.
Per engine
Run the Claude Code CLI on your own machine and drive it from any browser or phone: streamed transcripts, tool approvals and rendered diffs.
The OpenAI Codex CLI in a browser tab, with the rollout streamed live, diffs to review, and Claude Code chats in the same sidebar.
Grok Build on your machine, its sessions in the browser, and its permission prompts on your phone.
Terminals and editors
tmux keeps a session alive but shows you raw terminal output. Termdeck keeps every agent session alive and adds approvals, diffs, push and a board of every run.
Remote-SSH brings a whole editor to one remote box. Termdeck brings every Claude Code, Codex and Grok session on every machine into one tab.
Running several Claude Code sessions at once, across more than one machine, with approvals from your phone and takeover from any screen.
Phones and remote control
Remote Control hands one live Claude Code session to your phone. Termdeck shows every session on every machine, terminal ones included, plus Codex and Grok.
Happy puts sessions you start through its wrapper on your phone. Termdeck picks up every session on every machine, however it was started.
Agent view runs background sessions on one machine. Termdeck spans every machine and three engines, with push to your phone.
The web UI roundup puts these tools side by side. How-to guides walk through one job each, like running Claude Code from your phone, and the documentation covers every control.