Usage limits, handled
When an agent runs out, its reply says when it's back, and one @ gives the task to another.
Out of usage · back 15:40
All your coding agents.
One conversation.
Claude Code, Codex and the agents you already use, in one chat. Each works in its own copy of your project, and nothing reaches your code until you merge it.
Kin at work: Claude Code writes the fix in VS Code while Codex checks the app in Chrome.
Works with the agents you already have, on your own sign-ins
Claude Code
Codex
Grok Build
Cursor
Antigravity (Gemini)
Muse Code
Claude Code and Codex are tested end to end; Grok Build, Cursor, Antigravity and Muse Code are in beta. No API keys, and no per-use bills from us.
Steer without stopping
Type while an agent works. Claude Code takes your message on the way and answers everything in one reply; Codex gets it the moment it finishes.
Every agent in its own copy
Each conversation works in a git worktree on its own branch, so agents never touch your files or each other's. Read every change line by line, then merge in one click.
Share the work
Give a big job to one agent and it splits it with the agents that are free. Each takes its own files, they work at the same time, and the one that split it reviews everything before you merge it all at once.
See what they build, beside the chat
Start your app and Kin finds it. The page opens next to the conversation that's changing it, at desktop or phone width. When an agent browses the web, its browser plays live in the same panel.
Your computer, your rules
Agents start on edits only. When one wants more, Kin shows the exact command it asked for, and Allow and continue grants just that. Nothing merges by itself, and Stop means stop.
Your team at the same table
On a Team plan, everyone's Kin shares one space: a chat, conversations you hand each other, and what every agent is working on. Encrypted on your computers with a key only your team holds.
Built for a long day of work. The things you'd reach for in each agent's own app, in one place.
Side questions
/btw asks an agent something on the side. Its task goes on untouched.
From your phone
Scan a code in Settings and steer Kin from your phone: read, write and approve.
Reminders and schedules
“Every weekday at 8, run the tests.” The agent puts a card in its reply; nothing is set until you press Add.
Talk instead of typing
Your words appear as you speak, worked out on your own computer.
All your computers
Turn on sync and conversations follow you, encrypted with a key only your computers hold.
Commands, remembered
/model /effort /access /agent per agent, the way you left them.
Projects from GitHub
Open a repository, its issues and pull requests, and see CI right in the conversation.
A guide inside
Every setting, command and shortcut, one key away: F1.
Simple plans. Your agents run on your own subscriptions, so you never pay twice for the same model. Kin is free while it's in testing.
Free
$0
To try Kin on your own projects.
- Two agents in a conversation
- One project at a time
- Each conversation in its own copy
- Steering, sub-agents and the browser
Pro Most popular
$15a month
For people who code with agents every day.
- Every agent you have, at one table
- Unlimited projects and conversations
- @everyone, agent commands and the model menu
- The usage ring across all your plans
Team
$20a seat, a month
For teams that share agents and rules.
- Everything in Pro
- The team space: chat, hand-offs, who's on what
- Encrypted with a key only your team holds
- Shared instructions for each project
Bought inside Kin, by card. In Israel, ₪49 for Pro and ₪65 a seat for Team, charged in shekels. Cancel any time; a full refund within 14 days.
Bring your agents
to one table.
Kin is in testing on Windows and macOS, by invite. Tell us which agents you use and we'll send you the installer.