Back Up and Review UI Prototypes with YAML and Git
Understand Vinyasa document YAML versus RVV project files, preserve stable IDs, export backups, and review meaningful prototype changes in Git.
Two YAML shapes with different jobs
The document workspace edits a document containing schemaVersion, elements, pages, and settings. Compact examples may contain just elements and receive defaults during validation. An exported .rvv file is a project envelope: id, name, lastModified, state containing the document, and viewState containing scrollLeft, scrollTop, and zoom.
Paste document YAML into the YAML view. Import a complete .rvv file through File > Import File. Renaming arbitrary YAML to .rvv does not turn it into a valid project. The MCP export_project tool creates the proper envelope without writing a file itself.
Make backups deliberate
Browser autosave is convenient but not durable archival storage. Export .rvv after a meaningful milestone, store it in your chosen repository or backup system, and check that you can import it before clearing browser data or changing devices. Unapplied YAML drafts and AI proposals live only in tab memory and are not part of project autosave.
SVG and PNG are useful for discussion and documentation, but they are not editable Vinyasa project backups. Browser-local image references may not travel with a text export; keep source image assets separately and verify image availability in the destination environment.
Review changes as structured text
Keep element IDs stable across revisions. Review label and position changes separately from broad layout restructuring, and avoid reformatting entire files during a small edit. Metadata such as lastModified and viewport movement may add noise to .rvv diffs; focus design review on the state subtree.
Git cannot decide whether two conflicting coordinate edits produce a usable layout. Resolve text conflicts, validate the document, and inspect the result in the editor. Vinyasa does not provide built-in Git synchronization, multiplayer merging, or repository permissions. These are workflows around exported files, not hidden cloud features.
Choose a handoff format for the recipient
Share .rvv for continued Vinyasa editing, SVG/PNG for visual discussion, and a written interaction brief for implementation. No Angular or React application source is generated by these exports. Teams needing design-system libraries and live collaboration should also evaluate Penpot or Figma. For reviewed agent edits, use MCP.