How to use Claude to build a faster Svelte checker
A developer used Claude Code and Codex CLI to write svelte-check-rs, retaining architecture and real-codebase testing. The source reports cold, warm, and iterative speed improvements on a SvelteKit monorepo.
news.ycombinator.com
Prompt to try
Editorial adaptation, not the source author’s exact prompt. Replace bracketed inputs and review tool permissions first.
Implement [Svelte checker] with Claude Code and Codex CLI under human architecture direction. Parse [Svelte version], transform to TSX, invoke tsgo, and map diagnostics to original files. Test [real repositories] and report cold, warm, and iterative timings separately.
Reach people exploring AI agents
See placements →What happened
The author reports building the checker in about two days and measured 2.3× cold, 30× warm, and 16× iterative speedups on the stated SvelteKit monorepo.
Reported by gmaster1440 on Hacker NewsHow the agent did it
- Directed architecture while agents implemented a custom Svelte parser and TSX transformer.
- Used standalone tsgo checks and source maps to report errors at Svelte locations.
- Tested against real codebases and benchmarked a 650-file SvelteKit monorepo.
Where a person stepped in
The author made architecture decisions and tested against real codebases, including a reported benchmark on a SvelteKit monorepo.
A person approved each step before the agent continued.
The author built svelte-check-rs, a Rust replacement for svelte-check for Svelte 5 projects, in approximately two days using Claude Code and Codex CLI. The named configurations were Opus 4.5 and GPT-5.2 with xhigh reasoning. The author handled architecture decisions and testing against real codebases while the models wrote the parser, TSX transformer, diagnostics engine, and command-line interface.
The checker parses Svelte files with a custom Rust parser, transforms them to TSX in parallel through Rayon, and runs Microsoft's native TypeScript compiler, tsgo. Source maps translate diagnostic locations back to the original Svelte files. Writing TSX files to disk and invoking a standalone compiler enables persistent .tsbuildinfo files for incremental checks.
The post reports benchmarks on a 650-file SvelteKit monorepo using an M4 Max: 17.5 seconds versus 39.6 seconds cold, 1.3 versus 39.4 warm, and 2.5 versus 39.8 for iterative checks. These are the author's reported measurements. The stated motivation was shortening verification delays for coding agents working on a growing production codebase. Who it's for: developers exploring agent-assisted compiler tooling where architecture decisions and testing can remain human responsibilities.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Hacker News · gmaster1440
Frequently asked questions
What did the human developer handle?
The author says they focused on architecture decisions and testing against real codebases, while Claude Code and Codex CLI wrote the implementation components.
Where were the reported speed comparisons measured?
The post reports a 650-file SvelteKit monorepo on an M4 Max: 17.5s versus 39.6s cold, 1.3s versus 39.4s warm, and 2.5s versus 39.8s iterative. These are source-reported results.
More details
About this account
Claude Code and Codex CLI wrote the parser, transformer, diagnostics engine, and CLI while the author directed architecture and verification.
Platform: Hacker News. Attribution: https://news.ycombinator.com/item?id=46478789. Prompt: Suggested (written by Agents In Use, not the author's). Sourced by the Agents In Use pipeline.
Credit: gmaster1440 on Hacker News
Source published 3 Jan 2026
Source checked 14 Oct 2026
Tools and topics
Claude Code · Codex CLI · Rust · Rayon · tsgo
Svelte · Rust · Type checking
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.