@famcfdp A apple ainda está revisando o aplicativo.
Porém consigo disponibilizar via TestFlight, pega seu convite aqui: hiveterm.com/mobile/
Precisa ser o e-mail que você usa no icloud.
Sim, isso é o Claude Code rodando num iPhone.
O HiveTerm mobile espelha seu desktop: os agentes continuam rodando na sua máquina, você acompanha e responde do celular.
Vittorio, all 4 shipped today in v0.32.0.
Workspaces are now a list in the sidebar: click to switch (Ctrl+Shift+] / [ also works), drag to reorder, right-click a project to move it to another workspace. Agent needs you in a workspace you're not looking at? You get a notification and an amber dot on it, so your badge idea is in there too.
We tried the bar above the main panel like you suggested but it stole too much terminal height, so it went to the sidebar instead.
Also new: you can split agents from two different projects in the same view.
Thanks for the feedback, this was a good list. Tell me what still feels off after you update.
You already ask one AI for advice. The problem: it agrees with you.
In HiveTerm you put two models on opposite sides and let a third judge. Stack decisions, architecture reviews, launch plans. The answer that survives a debate is worth more than the answer that came first.
Here: Codex moderating Claude vs Gemini.
Available in the next release.
Split panes across projects. Working on two projects at once?
Drag an agent from any project onto a pane and they share one view.
Each pane shows its project's name so you always know which repo an agent is acting on, and the split stays reachable from both projects' sidebars.
O Hive não vem com modelo nenhum embutido. Ele só roda os CLIs oficiais (Claude Code, Codex, Gemini CLI, OpenCode, Aider) em terminais de verdade. Cada agente usa o modelo que você escolher no próprio CLI: no Claude Code é /model (Opus, Sonnet etc), no Codex e Gemini igual. A versão depende do seu plano, e dá pra rodar todos lado a lado.
@stphnbsvrt ha, the manager-manager stack stops here. hive runs the actual agents (claude code, gemini, codex) as real processes, each in its own pane. it's the workspace itself, not a layer sitting on top of one.
Claude Code's sub-agents run in the dark. You never actually see them work.
In Hive every sub-agent gets its own pane. You watch all of them build at the same time.
not really. they're separate processes, so it's the same cost as running them in 4 terminals, which you'd do anyway. hive only adds painting the visible panes (webgl per pane, falls back to canvas). hidden tabs stay alive but don't repaint. so seeing them costs about the same as running them blind.
@degenpark_eth they don't share a buffer. each agent is its own process with its own pty stream piped to its own pane, so nothing interleaves. you read them side by side instead of untangling one scrollback. that's the difference from piping 4 agents into a single terminal.
I want the following in Codex, Cursor, and OpenCode...
1. Pinned Messages: Let me pin assistant messages to the sidebar for things I want to keep track of but am not ready to address yet. Render as a checklist & jump navigation.
2. Notes: Give me a scratchpad for thoughts while