Free and open source · for macOS
Every agent you run, in one window.
Say what you want done.
Codealong sees every coding agent on your Macs: Claude Code, Codex, OpenCode, Antigravity and Cursor, on every account you pay for. Type a prompt and it goes where it belongs, to the agent already on it or to a free worktree on the account with room. Each agent keeps running in its own harness.
/plugin marketplace add codealong/codealong
9 windows · 3 sticky notes · no idea which one needs you
● Routing
Type it once. It goes where it should.
Other tools make you pick an agent before you type. Codealong starts with the prompt. Press ⌥⌘/ in any app and say what you want. It works out which session, worktree and account the work belongs to, and shows how sure it is.

A follow-up for the agent load-testing the limiter goes to that agent, not to a new one. ⏎ sends it. A new worktree is one ↓ away.
The agent already on it
Sent the way its harness takes a prompt: typed into its own terminal tab, through Claude Code’s background service, or as a new turn in the Codex app.
A worktree that’s free
New work lands in a reusable worktree with nothing in it. A brand-new worktree only in repos where that’s how you work.
A todo, for later
Not now? It’s kept as a todo, linked to the worktree it belongs to, ready to start when you are.
A chat of its own
A question that isn’t about your code gets a chat in no project, not a seat in one of your worktrees.

It checks before it starts.
When an agent that’s already running may be doing the same work, the start row says so and offers to send it there instead. Two agents never quietly race each other.

Auto picks the account with room.
Your starred accounts first, while they have room. An account at 90% of a limit has none. Of the rest, the one whose week resets soonest, so no week’s room goes unused. If a model’s own week runs low, it starts on the next model down.
Or decide it yourself, from the same box
- ⌘DMac any linked Mac
- ⌘TWorktree or a new one
- ⌘OOpens in terminal tab, background, Claude app, Codex app, or Codealong
- ⌘LAgent account, model and effort
Out of room mid-task?
A session stopped by a usage limit moves to an account with room and carries on. Codealong can suggest the move or make it.
One owner per todo.
A second agent started on the same todo is told to send its findings to the first, instead of redoing the work.
Change course between turns.
Switch a running Claude Code session’s model from its reply box, without starting over.
Suggestions come from Jev, TypeSafe’s small routing model, on your own key. It’s off until you turn it on, and Settings lists what it sends: what you type, plus session titles, branches and latest lines. Search, Auto and every override work without it. How routing works →
● Answering
Answer any agent without leaving what you’re doing.

Allow, Always or Deny.
Or deny it and say what to do instead, in the same keystroke, and the turn keeps going.
Nothing to switch on.
Claude Code’s own prompt stays up too, and whichever answer comes first wins. With Codealong closed, your agents ask you the usual way.
Only you approve.
Agents can see what another agent waits on, and can deny it. Only you can allow it.
Approvals and questions are answered for Claude Code (in its terminal tab, as a background job, or as a subagent), OpenCode, and any agent you run in Codealong. Codex, Cursor and Antigravity sessions show what they’re waiting on and open where they ask. What works with each harness →
● Every Mac
Start it on the Studio. Answer it from the laptop.
Link your Macs once. Each one shows the others’ worktrees and sessions beside its own, and says which Mac each one is on. Read a conversation running on another Mac, send it a message, answer its approval or start work there.

Peer to peer.
Your Macs find each other on your tailnet, or reach each other anywhere through iroh. They talk directly, encrypted end to end. There’s no Codealong server in the middle.
Link once.
On your tailnet, both Macs show the same six digits and you say yes on the other one. Anywhere else, enter a link code from one Mac on the other.
You decide what reaches in.
Remote control is off until you turn it on at that Mac. Once it’s on, your linked Macs can send prompts, answer approvals and start sessions there, and every request goes into a log.
Asleep isn’t gone.
A Mac that’s asleep or offline keeps its place on the board, showing its work as it was when you last saw it.
● Codealong for iPhone
Your agents, in your pocket.
Read any session on any of your Macs, and when one asks, answer it from the couch. It’s the same queue as the Mac: answer one and the next that waits on you is a tap away.
Every Mac, one list.
Sessions from each of your Macs, in the order the Mac shows them: what needs you first, then finished turns.
Answer from anywhere.
Allow or deny a call, answer its questions, reply, stop a turn, or start new work on the Mac, worktree and account you pick.
Peer to peer.
The phone links to your Macs the way they link to each other, end to end encrypted. No Codealong server, no account.
Free and open source, like the rest.
Other tools put their phone app on a paid plan. Codealong’s is free, Apache 2.0, with every feature.
It’s early: for now you install it from source with Xcode, and there are no notifications yet. A phone acts on a Mac only where that Mac has remote control on, and every action is logged there. Questions


And everything around the work.
See every featureStatus that reads the turn
Asks you, stuck or just slow, landed in main or not committed: the board says what you’d open the session to find out.
A canvas for your work
Every repo a frame, every worktree a column. Zoomed out, each card is a block in its status color.
Todos agents can work
Hand a todo to an agent and it stays linked to that agent, with one owner, dependencies and waits.
Pull requests, with their agents
Every PR joined to the worktree and sessions behind it. Approve, merge, or hand the review to an agent.
A board your agents can read
The codealong command and skill let agents see what others did, pick the next todo and check which account has room.
Built on the real harnesses
Claude Code, Codex, OpenCode, Antigravity and Cursor stay as their vendors ship them. Nothing wrapped, nothing proxied.
● Questions
Questions, answered.
Is Codealong really free?
Yes. The Mac app is free and open source under the Apache 2.0 license, with no account, no trial and no paid tier, and so is the iPhone app. A cloud service for teams is planned for later, and everything the apps do today stays free. More on pricing
Which agents does it work with?
Claude Code (in a terminal or as background jobs, on any number of accounts), the Claude app, the Codex CLI and app, OpenCode, Antigravity and Cursor. Cursor support is new and mostly read-only.
What Codealong can do with each depends on what that harness lets another program do. See what works with each harness
Do I have to start my agents from Codealong?
No. Codealong finds every session on your Mac, wherever it started: your terminal tabs, background jobs, the Claude and Codex apps. Keep starting agents the way you do now, and use the quick panel when it’s quicker.
Does it wrap or proxy my agents?
No. Codealong reads the files each harness already writes, and starts new work in the official harness, in your terminal or the vendor’s own app. Nothing sits between your agent and its model, and no API keys are involved. The optional plugin adds one hook to Claude Code so you can answer its permission prompts from Codealong. If you choose to run an agent in Codealong’s own panel instead, it talks to the agent over the Agent Client Protocol.
How does it decide where a prompt goes?
With suggestions on, Jev, a small routing model from TypeSafe, weighs what you typed against your running sessions and worktrees and suggests up to three places, each with how likely it is. Auto then picks the account: starred accounts first, never one at 90% of a limit, and of the rest the one whose week resets soonest. You can override the Mac, worktree, place and account from the same box. How routing works
What leaves my Mac?
Nothing, unless you turn something on. Transcripts stay on the Mac that ran them. Jev features are off until you add your own TypeSafe key and switch one on, and each switch lists what it sends. Linked Macs and your iPhone talk to each other directly, end to end encrypted, with no Codealong server. GitHub is read through your own gh sign-in.
Can I use several Claude accounts?
Yes. Each account runs the official Claude Code in its own config folder (CLAUDE_CONFIG_DIR), the way Anthropic documents for running accounts side by side, and Codealong finds them on its own, even from an alias in your shell. Follow each provider’s terms for how you use your accounts.
Can I approve agents from my phone?
Yes, with Codealong for iPhone. It links to your Macs peer to peer and lists every session on them. Where a Mac has remote control on, you can allow or deny calls, answer questions, reply and start sessions. It’s free, open source and early: for now you install it from source with Xcode, and it has no notifications yet.
Does it touch my git repos?
Codealong runs git read-only: it never commits, checks out, fetches or deletes. When work needs a new worktree, Claude Code makes it. Merging a pull request happens only when you press Merge.
Which terminals and Macs does it support?
macOS 13 or later. Ghostty is the best supported terminal; iTerm2, Terminal and tmux work too. To open a worktree, it knows VS Code, Cursor, Windsurf, Zed, Sublime Text, Xcode and JetBrains IDEs, among others.
How is it different from Conductor or Vibe Island?
Conductor and similar apps run agents inside their own window, so they see only what they launched. Notch apps like Vibe Island install hooks to show status and approvals. Codealong reads the agents you already run, across harnesses, accounts and Macs, and routes new work to the right one. See the full comparison




