Cloud Agent Development Environment

Cursor unveiled new tools for configuring and managing development environments for Cloud Agent. To run multiple agents in parallel while still letting them write code, run tests, call APIs, and access build systems all the way through, the environment itself needs to be sufficiently equipped.

• It supports multi-repository environments, letting you bundle multiple repos into a single environment. For organizations with interlinked repos, like microservices setups, agents can see the broader scope of a change's impact and complete edits, tests, and verification across repos.
• Dockerfile-based configuration has also been strengthened. Build secrets allow secure access to private package registries, and they apply only to the build step, never passed into the running agent environment. Improved layer caching makes builds 70% faster when the cache hits. The Dockerfile auto-generation feature is available in private beta and will roll out to Enterprise teams over the coming weeks.
• During setup, the agent asks questions, finds missing credentials, and displays the environment version. If setup fails, it falls back to the base image and shows a warning, so the agent doesn't stop immediately.
• Each environment has version history and rollback, and admins can restrict rollback to admins only. An audit log records the history of all changes, and egress and secrets are also isolated per environment.

Going forward, the goal is for environment configuration to evolve more autonomously in step with changes to the codebase.