Build C4 diagrams that survive review.

C4 describes architecture at four levels: Context, Container, Component, and Code. Fourlayer focuses on the first three as views over one connected model.

The C4 model is a way to describe software architecture at four levels of detail, created by Simon Brown. Each level answers a different question for a different audience. Context shows the system with the people and external systems around it. Container breaks that system into the applications, APIs, and data stores that actually run it. Component opens a single container into its internal parts. Code, the fourth level, is normally left to the IDE rather than drawn by hand. The point is that each level is a zoom of the same model, not a separate drawing, so a change made once stays true everywhere it appears. Most architecture diagrams fail review because they mix these levels on one canvas, so the reader cannot tell which question the picture is answering. Fourlayer keeps them as separate views over one model. The definitive reference is c4model.com.

Export the active view as Mermaid or Structurizr DSL for docs and repositories. You can also paste supported C4 code back into Fourlayer and keep editing it.

Fourlayer generates text from the model you draw, and reads text back into it. Every view exports as Mermaid, for docs, wikis, and pull requests where a diagram should live beside the code it describes, or as Structurizr DSL, the format for treating architecture as a versioned file in the repository. The generated text reflects only the elements visible in the active view, so a Container export does not leak Component detail. The direction reverses too: paste supported C4 code into Fourlayer and it becomes an editable model rather than a static import. That matters because architecture diagrams usually die at the export step. The image ships, the source is lost, and the next change starts from scratch. When the text and the canvas are the same model, the diagram survives the next review.

Context, Container, and Component views share one semantic model. Workspaces are private by default, with owner, editor, and viewer access.

Fourlayer supports three of the four C4 levels: Context, Container, and Component. All three are views over a single semantic model, so a system renamed in Context is renamed everywhere it appears. The available elements are people, software systems, external systems, containers, databases, components, and the relationships between them, plus an optional editable system boundary. Deployment views are not implemented yet; they are the next C4 level on the roadmap. Neither are billing, GitHub sync, public share links, realtime collaboration, or PDF export. Fourlayer is in free public beta, and those remain unbuilt rather than locked behind a plan. What does ship is export to Mermaid, Structurizr DSL, PNG at 1x, 2x, and 4x, and SVG, plus architecture health checks that link straight to the field that needs fixing. Workspaces are private by default and require a Google account, with owner, editor, and viewer roles for sharing.

Export the active view as Mermaid, Structurizr DSL, PNG, or SVG. The model stays editable for the next change.

Every view exports as text or as an image. The text options are Mermaid, which drops into docs, wikis, and pull requests where GitHub and GitLab render it natively, and Structurizr DSL, which treats architecture as a versioned file beside the code. Both reflect only the elements visible in the active view, so a Container export never leaks Component detail. The image options are PNG at 1x, 2x, and 4x, with an optional transparent background, and SVG when the diagram has to stay sharp at any size or be edited in a vector tool. Image exports carry a minimal Fourlayer watermark during the public beta. None of it is one-way: supported Mermaid and Structurizr code pasted back into Fourlayer becomes an editable model again, so the diagram survives the next change instead of dying at the export step.