Coming soon KataJam frontend katas
Micro-frontends and five rendering strategies, each measured against the same tests
Experiment
KataJam frontend katas
Micro-frontends and five rendering strategies, each measured against the same tests
The code lives in a private repository; this sheet and the demo, when there is one, show the work.
- Status
- Coming soon
- Type
- Experiment
- Running time
- November 2026 – November 2026
- Role
- Architect and sole engineer
- Domains
- frontend, architecture, performance, accessibility
- Cast
- React, Svelte, Module Federation, Rsbuild, Astro, Nextjs, TypeScript, Playwright, Lighthouse
Context and problem
Teams pick micro-frontends for independent deploys, and a rendering strategy by fashion, without knowing the price. Assumption: the same app built each way, with the same data and tests, shows what each choice wins and loses.
Screens
How it was built
Built in 6.1 h across 2 repos · USD 98.90 of AI · 318 tests · 8 audits
Source: Trysac/l21-kata-micro-frontends docs/development-report.md (lines 12, 15, 19): 2.7 h, USD 47.76, 134 tests, 4 audits; Trysac/l22-kata-render-ssr-ssg docs/development-report.md (lines 12, 15, 19): 3.4 h, USD 51.14, 184 tests, 4 audits, dated 21 November 2026.
Results with dated evidence
Outcomes as stated in the manifest; the measured figures are the ones with a date.
- L21: a shell and five remotes (one in Svelte) compose at run time from a versioned registry; the same 19 user flows pass on the micro-frontends and on a modular SPA.
- L21: a broken or tampered remote shows an accessible fallback while the rest keeps working.
- L22: one gallery as CSR, streaming SSR, SSG, ISR and islands, matching within 1 % of pixels; the same 12 flows pass on every variant.
- L22: a decision matrix built from the numbers, with when not to use each strategy.
Case study in preparation
The full case study adds these sections when it is published:
- Constraints
- Architecture diagrams (C4 context and containers)
- Three key decisions
- What went wrong
- Demo video
- What I would do differently
Highlights
module federation, runtime composition, rendering strategies, web vitals