svelte-core-bestpractices/examples/README.md

Independent synthetic Svelte experiment

From the extracted svelte-core-bestpractices directory, use compatible Node and:

npm ci --ignore-scripts --no-audit --no-fund
node examples/build_fixture.mjs
node examples/serve_fixture.mjs

Open http://127.0.0.1:5851/ in your normal browser. Click Run local observations once on a fresh page, inspect the displayed JSON and stop the server. Reload before repeating: checks mutate state and list contents. This server binds only to loopback and serves a strict allowlist. Dependencies come from the free public npm registry; binaries, node_modules and generated bundles are not bundled.

The builder writes a new compiler-evidence.json. The included browser-evidence is a historical Chrome 154 run on Windows; it is not rerun by the builder. Recorded versions: Node 24.10.0, Svelte 5.57.2, esbuild 0.28.2. Fifteen compiler/ server observations and 23 DOM observations passed. Eight original blocks are processed unchanged: four intentionally rejected contrasts/placeholders, four positive blocks with synthetic context where required. Hashes and wrappers are recorded. Local DOM events are programmatic, not trusted-input evidence.

BB Skills authored the runner, browser checks and synthetic components. One prop component deliberately stays stale and emits a compiler warning. Do not deploy the teaching fixtures. Node identity tests are not performance benchmarks. No hydration, SvelteKit, context isolation across users, paid API, real identity, database, comprehensive security or complete AI-client skill execution was tested. The snapshot is fixed; compare your own installed library version.