# 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]]