How it works
If you want to understand why diagramkit is structured the way it is — or extend it with a new engine — read this series.
| Page | What it covers |
|---|---|
| Browser pool | Single Chromium, four pages, reference counting, idle timeout. |
| Manifest system | SHA-256 hashing, stale detection, orphan cleanup, format accumulation. |
| Rendering pipeline | SVG-first, per-engine lane scheduling, atomic writes, raster conversion. |
| Color processing | WCAG luminance adjustment for dark SVGs. |
For consumer-facing guides (not internals), see the Guide → Architecture page.