# Cloud Development Environments (CDEs) A Cloud Development Environment is a **development workspace defined as code and provisioned on demand** on remote infrastructure: editor connects in, environment is disposable, "works on my machine" stops being a sentence anyone says. The laptop becomes a thin client; the environment becomes an artifact you version, template, and delete. ## The landscape | Product | Model | |---|---| | [[GitHub Codespaces]] | Managed, GitHub-native, [[Devcontainers]]-based | | [[Coder]] | Self-hosted, Terraform-templated, enterprise governance | | Gitpod / Ona | Managed, automation-first, drifted toward agent execution | | DevPod | Client-only open source, run devcontainers on any backend | The common substrate is usually [[Devcontainers]] (the spec that describes what an environment contains), with each product supplying provisioning, lifecycle, and access control around it. ## Why the category got a second life CDEs were a convenience for humans: onboarding in minutes, standardized toolchains, security by keeping source off laptops. Then [[AI Agents]] arrived and changed the economics. One developer needs one environment; a fleet of agents needs dozens: sandboxed, auditable, disposable. Every serious agent-infrastructure story now includes CDE machinery: [[Coder]] pivoted its whole positioning to governed agent workspaces, [[Claude Code Web]] runs sessions in cloud environments, and sandbox layers like [[Docker Sandboxes]] apply the same isolation idea at local scale. The pattern to remember: **environment-as-code was written for people, and agents turned out to be its heaviest users.** ## Related - [[Devcontainers]]: the definition format underneath most CDEs - [[GitHub Codespaces]] · [[Coder]]: the two poles (managed vs self-hosted) - [[Docker Sandboxes]] · [[WSL Container]]: local siblings of the isolation idea - [[Claude Code Web]]: agent sessions as cloud environments - [[AI Agents]]