# Cindy

The open-source AI agent Cindy connects a work computer, home desktop, laptop and phone to the same session, so a user can open a session on another device, continue the conversation and reach its files without copying everything into a new chat. It can also use model providers configured on another computer, including supported subscriptions, API providers and local models, and a built-in remote desktop on the phone lets the user operate a connected computer. The project states that remote access requires the machines to stay online, the same account sign-in and the relevant permissions, and that project files stay on their computer rather than being whole-disk synced.

The open-source AI agent Cindy connects a work computer, home desktop, laptop and phone to the same session, so a user can open a session on another device, continue the conversation and reach its files without copying everything into a new chat. It can also use model providers configured on another computer, including supported subscriptions, API providers and local models, and a built-in remote desktop on the phone lets the user operate a connected computer. The project states that remote access requires the machines to stay online, the same account sign-in and the relevant permissions, and that project files stay on their computer rather than being whole-disk synced.

The open-source client Cindy brings several harnesses and models, starting with Claude Code and Codex, into one AI agent that runs locally on the user's machine with real files and logged-in apps. Previously, changing harness or model usually meant rebuilding the workspace, memory and tools; Cindy keeps them continuous across switches, and one task can be planned, executed in parallel and reviewed by agents on different harness-model combinations. The client is Apache-2.0; the backend lives in a separate repository.

The open-source client Cindy brings several harnesses and models, starting with Claude Code and Codex, into one AI agent that runs locally on the user's machine with real files and logged-in apps. Previously, changing harness or model usually meant rebuilding the workspace, memory and tools; Cindy keeps them continuous across switches, and one task can be planned, executed in parallel and reviewed by agents on different harness-model combinations. The client is Apache-2.0; the backend lives in a separate repository.