AI 程式開發工具
Cursor Cloud Agents Can Start from a Blank Project, Moving Repositories, Previews, and Deployment into a Fully Cloud-Based Workflow
Cursor no longer requires users to connect an existing GitHub or other source code repository first. Agents can create projects directly and operate live environments through the browser. The new workflow shortens the path from prototype to deployment, while expanding the audit scope for cloud execution, network exposure, and software supply chain permissions.

Cursor has updated [Cloud Agents](https://cursor.com/changelog) with a “Start from scratch” option. Users can now create a new project with a prompt without first selecting a GitHub repository or another third-party SCM repository. The agent generates files, installs dependencies, and runs the application in Cursor’s cloud development environment. After reviewing the result, users can create a Cursor Origin repository with private or organization-internal visibility.
The technical significance of this change goes beyond removing the repo picker. Cursor forwards service ports from the agent environment to the browser, allowing users to interact directly with a Web application while it is being generated and specify UI changes through Design Mode. When connected to Vercel, the application can also be published to a public URL from the same workflow. The coding agent, previously centered on “clone a repository, create a branch, and open a PR,” has therefore expanded into a complete control plane covering “prompt, execute, interactively validate, create a repository, and deploy.” A [community announcement post](https://www.reddit.com/r/AIGuild/comments/1w0ao2p/cursor_cloud_agents_can_now_start_coding_from/) likewise focuses on this path that requires no existing repository.
For engineering teams, blank projects can reduce the setup cost of exploratory prototypes, but they also remove the default guardrails provided by existing repository templates, CI rules, branch protection, and dependency scanning. If an agent independently selects packages, runs installation scripts, and exposes services externally, risks can emerge before the first commit. Production adoption should therefore apply network egress allowlists, secrets management, dependency locking, build records, and deployment approvals during project creation, rather than adding them only after the repository has been created.
Documentation consistency is also worth watching. The current Cloud Agents documentation still primarily describes the traditional workflow of connecting an SCM, cloning a repository, and pushing a branch. It does not yet fully explain retention periods, backups, audit events, or permission inheritance for repo-less projects. Cursor has also not published separate quality or security benchmarks for this update. The next areas to monitor are whether the Origin API provides templates and policy-enforcement mechanisms, as well as the authentication, exposure scope, and idle-shutdown rules for browser port forwarding.