Saving a project

Click File → Save in the menu bar (or Ctrl+S). The first time you save, a file dialog appears where you choose the location and filename. The project is written as a single .tesseract.json file. The project title is used as the suggested filename.

On subsequent saves, Tesseract reuses the last file path — no picker dialog appears, so saving is instant. To choose a different location, use File → Save As (Ctrl+Shift+S), which always opens the file picker.

The file contains your entire architecture: all components, connections, subgraphs, flows, layer definitions, and project metadata (title, description).

The .tesseract.json format is plain JSON. You can version-control it with Git, diff it, or generate it programmatically.

Loading a project

Click File → Load in the toolbar. A file picker opens where you select a .tesseract.json file. The current architecture is replaced by the loaded one, and the view automatically fits all components.

If you have unsaved changes, a confirmation dialog asks whether you want to discard them before loading.

Creating a new project

Click File → New in the toolbar. You are offered two options:

As with loading, you will be asked to confirm if there are unsaved changes.

Auto-save

Tesseract automatically saves your work to the browser's local storage as you edit. If the application closes unexpectedly — a crash, a power loss, or an accidental tab close — your latest state is preserved.

On the next launch, the auto-saved state is restored automatically. This is a safety net, not a replacement for saving to a file: local storage can be cleared by the browser, so always save important work to disk.

Unsaved changes

When you have changes that haven't been saved to a file, the window title shows an indicator. Before any operation that would discard your work (New, Load), a confirmation dialog appears so you don't lose anything by accident.

Mermaid import/export

Tesseract can export and import Mermaid flowchart diagrams, making it easy to embed your architecture in Markdown files, READMEs, wikis, or any tool that supports Mermaid rendering.

Exporting: click File → Export Mermaid. The entire architecture is serialized as a Mermaid graph TD flowchart with layer subgraphs, component nodes, protocol-labeled edges, and nested subgraphs. The output is saved as a .mermaid file.

Importing: click File → Import Mermaid. Select a .mermaid or .mmd file. The parser reconstructs components, connections, and layers from the Mermaid source and replaces the current architecture. If you have unsaved changes, a confirmation dialog appears first.

The exported file includes special metadata comments (@layer, @component, @connection, @flow) that preserve component types, technologies, positions, descriptions, and data flows. When re-importing, all this metadata is restored for a lossless roundtrip. Plain Mermaid files without metadata are also supported with sensible defaults.

The MCP tools export_mermaid and import_mermaid provide the same functionality programmatically, so Claude can export or import Mermaid diagrams directly from your terminal. See MCP API Reference.

Sharing a project

To share an architecture with a colleague:

  1. Save the project to a .tesseract.json file
  2. Send the file (email, Slack, Git, shared drive…)
  3. The recipient opens it with File → Load

The file is self-contained — no external dependencies. Anyone with Tesseract can open it. However, projects saved with a Pro account may use features that are not available on the Free tier (custom shapes, additional layers, flows). A Free-tier user can still open the file, but some elements may not render the same way.

Since the file is plain JSON, it works well with version control. Commit it alongside your source code to keep architecture and implementation in sync.