Rule assets
132+Corpus scale drives the complexity of the publication system.
Performance Whitepaper
Performance is no longer treated as raw build time alone. For a research-grade publication surface, the meaningful questions are whether the generation pipeline is stable, the reading path is efficient, the figures are reliable, and the shell remains maintainable.
Performance Signals
Expand performance beyond build time into maintainability and evidence reliability
This chapter frames performance across four dimensions: generation, navigation, figure reliability, and maintainability.
Domains
32+Breadth requires a more disciplined IA and snapshot layer.
Phases
3Curation, orchestration, and publication define the generation pipeline.
Generation pipeline
The generation pipeline has three stages:
- Asset discovery across
rules/ - Snapshot generation for
categoryDistribution,coverageBuckets,qualitySignals, andtimelineSignals - Publication rendering through VitePress
The main benefit is alignment: the site shell stays grounded in repository facts instead of drifting into static copy.
Maintainability
Maintainability is one of the core performance metrics in this redesign.
- Single data base: metrics come from one snapshot layer
- Single visual language: the whitepaper components share the same figure tokens
- Single reading path: introduction, architecture, algorithms, performance, references, and evidence remain predictable
That matters more in the long run than shaving a few seconds from a one-off build.
Figure reliability
Figure performance here means:
- light and dark mode stay readable
- labels, lines, and surfaces come from consistent tokens
- SVG diagrams remain visible instead of relying on brittle theme hacks
Why this chapter belongs in the whitepaper
Senior readers want to know whether the generation pipeline is trustworthy, whether the shell will remain maintainable, and whether the figures and navigation scale as the corpus grows. Those are whitepaper questions, not implementation trivia.