How Ideagram Works
Ideagram keeps architecture knowledge in a project. Pages explain the decision, diagrams make it visual, and MCP lets compatible tools read and update the same context. The diagram below is rendered by the portable Ideagram Viewer — it is a real, read-only Ideagram diagram, not an image.
Loading the interactive Ideagram viewer…
Use the controls to zoom and pan. The viewer accepts Mermaid source and renders it through Ideagram's portable graph pipeline, the same rendering surface used for embedded diagrams.
The flow
| Part | What it owns today |
|---|---|
| Project | The shared boundary for diagrams, Pages, notes, folders, and project context. |
| Pages and notes | Markdown knowledge: decisions, specifications, links, and diagram embeds. |
| Diagrams | Their own components and connections. Diagrams do not yet share one global model. |
| Optional model | Reusable model entities can be linked from diagram components; this is additive, not a replacement for a diagram's visual data. |
| MCP client | Reads and updates the project context through supported MCP tools. |
A practical loop
- Start with a Project for the system or initiative you are describing.
- Capture the decision in a Page or Note.
- Create a Diagram and embed or reference it from that Page with a wiki link.
- Use an MCP client when you want an agent or developer tool to inspect the same project context.
This is the current alpha model. Shared, model-first architecture flows remain roadmap work; they are not represented as a global shared canvas today.