Use it as your own browser. Let AI agents work alongside you, in tabs of their own, with your sign-ins and under your rules.
One named group of tabs for each thing you're doing, and one for each task an agent does. Close a space and pick it up later.
A live grid of every space, refreshed every second while an agent works, with what it did last.
An agent's tab is framed, with a cursor where it's working. Take control or stop the agent with one click. The agent can't see or press those buttons.
Keep what an agent did, watch it back step by step, or run it again as an automation. Off until you turn it on.
Agents use the sessions in your window. Passwords, two-factor and payments stay your steps.
On macOS, an app of its own built on open-source Chromium, with everything it needs inside.
Claude Code, Codex, or any agent with a terminal. Teach it once, then ask it to use Kin Browser.
# once kin-browser skill install # what the agent runs kin-browser open https://github.com --space "Research" kin-browser click e42
Claude Desktop, Cursor, or any MCP client. Add it as an MCP server. Or call the local API from your own program.
{
"mcpServers": {
"kin-browser": {
"command": "kin-browser",
"args": ["mcp"]
}
}
}
On macOS, no: it's a browser of its own, built on open-source Chromium, the project Chrome is built from. On Windows and Linux it currently opens in your Chrome or Edge, with a profile of its own.
No. It has its own profile, empty until you use it.
No. It can read and click on public websites, in its own tabs. It can't run code, open files or reach your local network, and you can stop it at any time.
Only if you turn recording on. Each step is a small screenshot, kept on your computer only, under a limit you set (1 GB unless you change it). The oldest go first.
Yes. Kin Browser is open source under the MIT licence.