roadmap

optica is a static-site publisher for markdown knowledge graphs. the rendering pipeline (markdown → HTML → templates → CSS + minimal JS) is constrained by the web platform — DOM, Canvas, CSS. that ceiling is acceptable for documents up to ~10⁴ nodes.

high-performance interactive graph rendering at scale (10⁵+ nodes, GPU compute, realtime tri-kernel) is out of scope. it belongs to a standalone renderer in cyb that targets wgpu/Vulkan/Metal directly, sharing buffers with the layout and ranking compute. optica's graph view stays a read-only static visualization.

rendering — what this project owns

scope: improvements to the static rendering pipeline that fit within DOM/Canvas constraints.

tier 1 — small wins, high value

dark/light mode toggle (CSS variables already in place, runtime switcher missing)

responsive sweep — current style.css has 2 media queries; sidebar, peers, and minimap overflow on small viewports

code blocks: copy-to-clipboard button, optional line numbers

images: loading="lazy", srcset, captions from alt text, lightbox on click

accessibility pass: skip-to-content link, ARIA landmarks (nav, main, complementary), visible focus styles, heading hierarchy validation

minify style.css (1770 LOC) and graph.js (709 LOC) in release builds

tier 2 — feature gaps vs peers

mermaid diagram support (server-side render preferred)

KaTeX SSR or pre-rendered math (eliminate per-page DOM walk and flash)

code block extensions: tabbed groups, diff highlighting, language label

admonitions: allow nested markdown inside body, expand type set

search UI: typeahead with result snippets and focus-weighted scoring

inline minimap data (current per-page minimap re-fetches graph-data.json)

tier 3 — query and content

expand the static query language: graph traversal (neighbors, path), date filters, aggregates, count

footnote styling and back-references

redirect map for renamed pages (alias resolution covers in-graph, not external URLs)

rendering — out of scope

these belong to a standalone GPU renderer in cyb, not optica:

realtime force layout on 10⁵+ nodes

compute-shader tri-kernel running in the viewport

level-of-detail rendering keyed on focus / cyberank

frustum and focus culling for navigable mole-scale graphs

shared GPU buffers between layout, ranking, and rendering

WebGPU/native renderer with instanced node draws

optica's responsibility ends at producing static graph data (nodes, edges, focus, gravity) consumable by any renderer — including the future GPU renderer.

non-rendering tracks

other tracks (full-text search via tantivy, plugin hooks, themes-as-packages, exports, i18n) live outside this document until prioritized.