nyaomaru

In TypeScript, unknown is everywhere. JSON.parse, API responses, external inputs… But we still need to use the data safely. That means one thing: → we have to prove what it is So, I built a small library `is-kit`. GitHub: github.com/nyaomaru/is-...

Another thing I noticed comparing TypeScript validators 😸 “Exact object validation” sounds simple. But inherited fields, extra symbol keys, sparse arrays, or an extra key with `undefined` can make them disagree. JavaScript is weird. 😹 Full comparison tomorrow! 🦐

I’m comparing 7 TypeScript runtime validation approaches for my next article. One thing surprised me: “max length: 100” didn’t mean the same thing everywhere. 👀 For 51 😀: • 51 Unicode code points • 102 UTF-16 code units Same-looking rule, different runtime contract. 😹

TypeScript 7.1 Beta is interesting 👀 What caught me isn’t new type syntax, but the tooling side. `typescript/ast` is getting a proper public API, and Content Mappers make Vue, Svelte, Astro, etc. much easier to integrate. TypeScript 7.x is getting fun. 😸

changelog-bot v0.10.0 / v0.10.1 are out 🚀 v0.10.0: johnnylemonny added sub-change WHY mapping for multi-change PRs. v0.10.1: 5 new contributors joined! 🎉 KingEmma7 / jay23606 / Prashaxt / mangomac / erSpell Thanks for helping the project grow! 😸 github.com/nyaomaru/changelog-bot/r…

I've been comparing 7 approaches to runtime validation in TypeScript. 🤔 Native TS, is-kit, Runtypes, Valibot, Zod, ArkType, typia. Same `User` type. Same validation rules. But making them behave the same was harder than making them compile. 😹 Full comparison next Wednesday! 🦐

My favorite story from the is-kit 100⭐ article: An is-kit refinement using `unknown` exposed an inference edge case in Effect's `catchIf`, which was later fixed with `NoInfer`. 🚀 I love bugs that only appear when two libraries meet in real code. 😸 dev.to/nyaomaru/is-kit-reached-100…

YAY! is-kit reached 100 GitHub stars ⭐🎉🎉 A lot happened since 50 stars: new APIs, stricter guards, more real OSS usage, and even an Effect type inference bug discovered through an is-kit guard. 😸 I wrote about what changed between 50 → 100 👇 dev.to/nyaomaru/is-kit-reached-100…

TypeScript 7.1 Beta is now available! Now with a preview API, content mapper support, new features, and some great optimizations. Give it a try and give us some feedback!

is-kit v1.16.0 is out 🎉 New: discriminatedUnionWithFallback It lets you handle future/unknown discriminant values explicitly while keeping downstream code on a finite union. Known but malformed variants still fail validation 😸 https://github.com/nyaomaru/is-kit/releases/tag/v1.16.0 #TypeScript

changelog-bot v0.9.2 is out 🎉 Fixed a race in the release workflow by pinning changelog generation to the release version. And we welcomed a new contributor, Anantha-K! Thanks for the great first contribution 😸 https://github.com/nyaomaru/changelog-bot/releases/tag/v0.9.2 #TypeScript #OpenSource

changelog-bot v0.9.0 is out 🚀 Every source change must now be accounted for: included in the release or recorded with an explicit exclusion reason. No silently disappearing changes. Deterministic structure, AI enrichment 😸 https://github.com/nyaomaru/changelog-bot/releases/tag/v0.9.0

filter(Boolean) and “remove nullish values” are not the same thing. Boolean also removes 0, false, and ''. If you only mean “remove null and undefined”, say exactly that. 😸 https://is-kit.dev/guides/filter-null-and-undefined #TypeScript

I’m so used to typing `code .` in the terminal from years of using VS Code. But lately, when I want to use Codex, I keep typing `codex .` instead 😹 Please tell me I’m not the only one. 😹

unknown isn’t a problem to eliminate. It’s an honest description of a value you haven’t verified yet. 😸 Validate it without reaching for a schema by default: https://is-kit.dev/guides/validate-unknown-without-schema-library #TypeScript #OpenSource

Not like the lib itself caught it, but when I created a reusable operator: export const isNotFoundError = and( isObject, refineDefinedKey('reason', equalsKey('_tag', 'NotFound')), ) I noticed my streams having unknown in the error channel github.com/Effect-TS/ef...

unknown is an honest description of an unverified value. as User changes what TypeScript believes — not the runtime value. A small composable guard may be all you need. 🛡️ https://dev.to/nyaomaru/how-do-you-validate-unknown-in-typescript-without-a-schema-library-1a11 #TypeScript #JavaScript

I published a new TypeScript article ✍️ How do you safely turn `unknown` into something useful without a full schema library? I walk through type guards, `struct`, `safeParse`, JSON boundaries, exact object validation, and where schema libraries still make more sense. #TypeScript #OpenSource

I wrote my own function to replace nonnull assertions but you don’t have to. Use is-kit instead. You can’t rely on typescript at runtime. Sometimes it’s not a problem that can be solved with Zod.

`is-kit` passed 100 GitHub stars ⭐🎉 I started it as a small zero-dependency TypeScript type guard library, so it’s really nice to see it gradually reaching more developers. 😸 Thanks for all the feedback, issues, PRs, and stars!! github.com/nyaomaru/is-... #TypeScript #OpenSource

“0.8 feels safe” 😹 That’s exactly what I wanted to avoid with changelog-bot. John and I expanded the WHY benchmark from 14 to 50 cases, then swept thresholds and compared precision, recall, and F1. Guess less. Measure more. #TypeScript #OpenSource

Maybe WHY extraction isn't a generation problem. Maybe it's a selection problem. 🤔 Jev evaluates rationale already written in the PR. If the evidence is weak, nothing gets added. New article 👇 #TypeScript #OpenSource

unknown is doing its job. An API response doesn’t become a User just because our app expects one. 😸 The interesting part is how we establish trust at runtime and whether every boundary really needs a full schema library. #TypeScript #OpenSourc

Green build ≠ verified package. For my tsup → tsdown migration, I packed the npm tarball and tested it from a temporary consumer: ESM, CJS, and TypeScript. Test what users actually install. 🔧 dev.to/nyaomaru/mig... #TypeScript #OpenSource

I migrated is-kit from tsup to tsdown. 🔧 Tiny config change. Green build. Wrong package contract. 🙀 I wrote what changed and how I verified the package users actually install. dev.to/nyaomaru/mig... #TypeScript #OpenSource

Really happy to share that my TypeScript Compiler API article was featured in DEV’s Top 7 Posts of the Week! 😸🎉 I’m honestly really happy to be featured again!! Huge thanks to the DEV team for featuring it, and congrats to all the other authors too! 🚀 dev.to/devteam/top-... #TypeScript #OpenSource

For a library, the source code isn’t the final artifact. The packed npm package is. That became very obvious during my tsup → tsdown migration, so I tested the actual tarball in temporary ESM, CJS, and TypeScript consumers. Full write-up tomorrow. 🔧 #TypeScript #OpenSource

A small tsup → tsdown migration surprise. Before: index.mjs index.js index.d.ts First tsdown build: index.mjs index.cjs index.d.mts index.d.cts The build passed. Nothing was “wrong” with tsdown, but reasonable new defaults can still break an existing package contract. 👀 #TypeScript #OpenSource

What if AI didn’t write the “Why” in a CHANGELOG at all? In changelog-bot v0.7.0, Jev can select rationale already present in the PR description instead of generating new prose. Less invention. More grounding. Still experimental, but I like the direction. 👀 #OpenSource #AI #DevTools

A green build doesn’t always mean a safe release. 👀 I migrated a real TypeScript OSS library from tsup to tsdown. The config change was tiny. The build passed. But the generated package no longer matched its existing contract. Full write-up Wednesday. 🔧 #TypeScript #OpenSource #tsdown

changelog-bot v0.7.0 🚀 Experimental TypeSafe Jev support for WHY extraction is here. It selects rationale in PR descriptions instead of generating prose. Huge thanks to github.com/johnnylemonny! 🙌 github.com/nyaomaru/cha... #OpenSource #TypeScript #CHANGELOG #DevTools

Reusable AST guards can lose child-property narrowing. With is-kit, I compose existing ts.isX guards with a generic refineKey() helper while preserving parent + child narrowing. 🔧 dev.to/nyaomaru/typ...

I wrote new article! 🔧 Reusable parent + child narrowing gets awkward in the TypeScript Compiler API. I explored property refinement with `is-kit` and what changes in TypeScript 7. dev.to/nyaomaru/typ... #TypeScript #OpenSource

TypeScript 7 changed the Compiler API surface, but one thing surprised me: a broad `ast.Node` still isn’t a closed discriminated union. So `kind === CallExpression` doesn’t replace `isX` narrowing for that input. Full write-up tomorrow. 🔧 #TypeScript #CompilerAPI

🎉 Changelog Bot is now live in GitBot’s in-app Library! You can now discover and use @nyaomaru/changelog-bot directly from GitBot 😸 Huge thanks to Sahil and the GitBot team for adding it! gitbot.codeongrass.com/bots/changelog-bot #OpenSource #DevTools

What does `arguments[0]` actually prove? Nothing, if the array is empty. A reusable guard for the first argument needs to prove: 1. index 0 exists 2. the value matches 3. the parent type remembers both facts Tiny runtime check, surprisingly interesting type problem. #TypeScript

is-kit v1.15.0 is out! 🚀 A discussion about type-guard drift led to `discriminatedUnion`. It checks union coverage at compile time, so missing, extra, or mismatched branches can’t silently slip through. 👀 github.com/nyaomaru/is-... #TypeScript #OpenSource #is-kit

Optional AST children hide a different problem. `initializer?: Expression` isn’t just “another property to narrow.” A reusable guard must first prove that the property exists, then refine its value. That runtime contract deserves to be explicit. #TypeScript #OpenSource

TypeScript 5.5 added inferred type predicates, which is great. But a compound check like: `CallExpression → expression → Identifier` can still end up as plain `boolean` when extracted. So where exactly does reusable narrowing stop being “free”? 👀 #TypeScript

A type guard can prove both the parent and a child property at runtime. But if you extract the check into a reusable function, does TypeScript preserve both facts? `CallExpression → expression → Identifier` That small question turned into a bigger TypeScript rabbit hole. 😸 #TypeScript #WebDev

A thought from writing this article 😸 A stale type guard can accept bad data, and the bug may appear much later far from the boundary that let it through. 👀 That distance is what makes silent drift painful. dev.to/nyaomaru/you... #TypeScript #OpenSource #OSS

TypeScript article! 🔧 A handwritten type guard can stay valid while runtime checks fall behind the type. Why this drift happens, and how `typedStruct<User>()` catches it. 👀 dev.to/nyaomaru/you... #TypeScript #OpenSource

This week I’ve been sharing one TypeScript maintenance problem: A hand-written type guard can keep compiling even after the type it claims to validate has changed. Tomorrow I’ll publish the full write-up — including drift, optional fields, nested types, and where `typedStruct` fits. 🔧 #TypeScript