"Ona runs either in the company's multi-tenant cloud, with US and Europe regions, or on an AWS VPC (virtual private cloud) for those who require more control and isolation. The remote environments run Visual Studio Code in the browser, though it is also possible to open projects in a number of IDEs running locally, including VS Code, Cursor, Windsurf, JetBrains IDEs, and Zed. These IDEs require the Gitpod/Ona extension."
"Landgraf said that Ona's remote environments are sandboxed and ephemeral, which in his view enables "reliable, secure autonomy." Autonomy here means using agents in autonomous mode, where they are empowered to write code and perform tasks without manual approval. Running remotely reduces the risk of AI agents misbehaving on a local PC or business network, though given that Ona asks for full read/write access to GitHub repositories and workflows, for example, risks from attacks such as prompt injection remain."
Ona, formerly Gitpod, combines remote development environments with integrated AI agents and positions agents as the next era of development tools. The company reports agents co-authored 60 percent of PRs merged on main and contributed 72 percent of merged code in one week. Environments run Visual Studio Code in the browser and can be opened in local IDEs (VS Code, Cursor, Windsurf, JetBrains, Zed) using the Gitpod/Ona extension. Deployments run in a multi-tenant cloud (US and Europe) or in a customer AWS VPC. Environments are sandboxed and ephemeral to enable autonomous agent execution, with guardrails such as short-lived access, a command deny list, and audit logs. Agents can operate in autonomous, assisted, or manual modes.
Read at Theregister
Unable to calculate read time
Collection
[
|
...
]