Cognition Adds SSH and Cloud Handoffs to Devin CLI
Cognition has launched Devin Cloud, allowing developers to hand off coding tasks between local machines and remote virtual machines while directly accessing the AI agent's environment via SSH.

Cognition has released Devin Cloud in Terminal alongside a new devin ssh command, bridging the gap between local command-line interfaces, hosted sessions, and remote development virtual machines. Developers can now spin up, steer, and resume cloud-based agent sessions directly from their terminal using the /cloud command. The update introduces a /handoff command that seamlessly transfers active development branches between a local machine and a dedicated cloud VM in either direction, preserving the exact state of the branch.
By running devin ssh, developers gain direct shell access to the remote environment where the Devin agent is actively writing code and running tests. This hosted environment, which runs on Ubuntu 22.04 for Linux targets, can also be provisioned with macOS or Windows depending on the build requirements. This direct access allows practitioners to inspect source trees, review logs, transfer files using scp, and use port forwarding to test web applications running inside the VM.
The cloud sessions persist across multiple interfaces, meaning a developer can start a task in the terminal, monitor it via the web dashboard or Devin Desktop, and resume it later using the --resume flag. To encourage adoption, Cognition is offering free sessions utilizing its in-house SWE-2 coding model through October 8. Developers can install the updated command-line tool by running curl -fsSL https://cli.devin.ai/install.sh | bash in their terminal.
For software engineers, this release simplifies the process of managing long-running autonomous tasks. Instead of tying up local system resources for heavy builds or test suites, a developer can prototype locally, use /handoff to delegate the heavy lifting to a remote VM, and retrieve the completed branch for final review. However, teams should carefully evaluate how these handoffs handle complex Git states, dirty working trees, and environment-specific secrets before fully integrating the tool into their production pipelines.
This is our own summary of reporting by AlphaSignal



