TypeScript Guidelines

Core Rules

Always use type instead of interface in TypeScript.

readonly only for arrays and maps: Never use readonly on primitive properties or object properties. The modifier is shallow and provides little protection for non-collection types. Use it only where mutation is a realistic footgun:

// Good - readonly only on the array type Config = { version: number; vendor: string; items: readonly string[]; };

// Bad - readonly everywhere is noise type Config = { readonly version: number; readonly vendor: string; readonly items: readonly string[]; };

Exception: Match upstream library types exactly (e.g., standard-schema interfaces). See docs/articles/readonly-is-mostly-noise.md for rationale.

Acronyms in camelCase: Treat acronyms as single words, capitalizing only the first letter:

// Correct - acronyms as words parseUrl(); defineKv(); readJson(); customerId; httpClient;

// Incorrect - all-caps acronyms parseURL(); defineKV(); readJSON(); customerID; HTTPClient;

Exception: Match existing platform APIs (e.g., XMLHttpRequest ). See docs/articles/acronyms-in-camelcase.md for rationale.

TypeScript 5.5+ automatically infers type predicates in .filter() callbacks. Don't add manual type assertions:

// Good - TypeScript infers the narrowed type automatically const filtered = items.filter((x) => x !== undefined);

// Bad - unnecessary type predicate const filtered = items.filter( (x): x is NonNullable<typeof x> => x !== undefined, );

When moving components to new locations, always update relative imports to absolute imports (e.g., change import Component from '../Component.svelte' to import Component from '$lib/components/Component.svelte' )

When functions are only used in the return statement of a factory/creator function, use object method shorthand syntax instead of defining them separately. For example, instead of:

function myFunction() { const helper = () => { /* ... */ }; return { helper }; }

Use:

function myFunction() { return { helper() { /* ... */ }, }; }

Prefer factory functions over classes: Use function createX() { return { ... } } instead of class X { ... } . Closures provide structural privacy—everything above the return statement is private by position, everything inside it is the public API. Classes mix private /protected /public members in arbitrary order, forcing you to scan every member and check its modifier. See docs/articles/closures-are-better-privacy-than-keywords.md for rationale.

Switch Over If/Else for Value Comparison

When multiple if /else if branches compare the same variable against string literals (or other constant values), always use a switch statement instead. This applies to action types, status fields, file types, strategy names, or any discriminated value.

// Bad - if/else chain comparing the same variable if (change.action === 'add') { handleAdd(change); } else if (change.action === 'update') { handleUpdate(change); } else if (change.action === 'delete') { handleDelete(change); }

// Good - switch statement switch (change.action) { case 'add': handleAdd(change); break; case 'update': handleUpdate(change); break; case 'delete': handleDelete(change); break; }

Use fall-through for cases that share logic:

switch (change.action) { case 'add': case 'update': { applyChange(change); break; } case 'delete': { removeChange(change); break; } }

Use block scoping ({ } ) when a case declares variables with let or const .

When NOT to use switch: early returns for type narrowing are fine as sequential if statements. If each branch returns immediately and the checks are narrowing a union type for subsequent code, keep them as if guards.

See docs/articles/switch-over-if-else-for-value-comparison.md for rationale.

Record Lookup Over Nested Ternaries

When an expression maps a finite set of known values to outputs, use a satisfies Record lookup instead of nested ternaries. This is the expression-level counterpart to "Switch Over If/Else": switch handles statements with side effects, record lookup handles value mappings.

// Bad - nested ternary const tooltip = status === 'connected' ? 'Connected' : status === 'connecting' ? 'Connecting…' : 'Offline';

// Good - record lookup with exhaustive type checking const tooltip = ({ connected: 'Connected', connecting: 'Connecting…', offline: 'Offline', } satisfies Record<SyncStatus, string>)[status];

satisfies Record<SyncStatus, string> gives you compile-time exhaustiveness: if SyncStatus gains a fourth value, TypeScript errors because the record is missing a key. Nested ternaries silently fall through to the else branch.

as const is unnecessary here. satisfies already validates the shape and value types. as const would narrow values to literal types ('Connected' instead of string ), which adds no value when the output is just rendered or passed as a string.

When the record is used once, inline it. When it's shared or has 5+ entries, extract to a named constant.

See docs/articles/record-lookup-over-nested-ternaries.md for rationale.

Type Co-location Principles

Never Use Generic Type Buckets

Don't create generic type files like $lib/types/models.ts . This creates unclear dependencies and makes code harder to maintain.

Bad Pattern

// $lib/types/models.ts - Generic bucket for unrelated types export type LocalModelConfig = { ... }; export type UserModel = { ... }; export type SessionModel = { ... };

Good Pattern

// $lib/services/transcription/local/types.ts - Co-located with service export type LocalModelConfig = { ... };

// $lib/services/user/types.ts - Co-located with user service export type UserModel = { ... };

Co-location Rules

• Service-specific types: Place in [service-folder]/types.ts

• Component-specific types: Define directly in the component file

• Shared domain types: Place in the domain folder's types.ts

• Cross-domain types: Only if truly shared across multiple domains, place in $lib/types/[specific-name].ts

types.ts Is A Code Smell (Prefer Computed Types Over Manual Declarations)

When a type can be derived from a runtime value, derive it. Don't declare it manually in a separate file.

// Good — type is computed from the runtime definition export const BROWSER_TABLES = { devices, tabs, windows }; export type Tab = InferTableRow<typeof BROWSER_TABLES.tabs>;

// Good — type is derived from schema const userSchema = z.object({ id: z.string(), email: z.string() }); type User = z.infer<typeof userSchema>;

// Bad — manually declaring what already exists as a runtime value // types.ts export type Tab = { id: string; deviceId: string /* ... */ };

If every type in a types.ts can be derived with typeof , z.infer , InferTableRow , ReturnType , etc., the file is redundant. Put each type next to the runtime value it's computed from.

Constant Array Naming Conventions

Pattern Summary

Pattern Suffix Description Example

Simple values (source of truth) Plural noun with unit Raw values array BITRATES_KBPS , SAMPLE_RATES

Rich array (source of truth) Plural noun Contains all metadata PROVIDERS , RECORDING_MODE_OPTIONS

IDs only (for validation) _IDS

Derived from rich array PROVIDER_IDS

UI options {value, label}

_OPTIONS

For dropdowns/selects BITRATE_OPTIONS , SAMPLE_RATE_OPTIONS

Label map _TO_LABEL (singular) Record<Id, string>

LANGUAGES_TO_LABEL

When to Use Each Pattern

Pattern 1: Simple Values -> Derived Options

Use when the label can be computed from the value:

// constants/audio/bitrate.ts export const BITRATES_KBPS = ['16', '32', '64', '128'] as const;

export const BITRATE_OPTIONS = BITRATES_KBPS.map((bitrate) => ({ value: bitrate, label: ${bitrate} kbps, }));

Pattern 2: Simple Values + Metadata Object

Use when labels need richer information than the value alone:

// constants/audio/sample-rate.ts export const SAMPLE_RATES = ['16000', '44100', '48000'] as const;

const SAMPLE_RATE_METADATA: Record< SampleRate, { shortLabel: string; description: string }

= { '16000': { shortLabel: '16 kHz', description: 'Optimized for speech' }, '44100': { shortLabel: '44.1 kHz', description: 'CD quality' }, '48000': { shortLabel: '48 kHz', description: 'Studio quality' }, };

export const SAMPLE_RATE_OPTIONS = SAMPLE_RATES.map((rate) => ({ value: rate, label: ${SAMPLE_RATE_METADATA[rate].shortLabel} - ${SAMPLE_RATE_METADATA[rate].description}, }));

Pattern 3: Rich Array as Source of Truth

Use when options have extra fields beyond value /label (e.g., icon , desktopOnly ):

// constants/audio/recording-modes.ts export const RECORDING_MODES = ['manual', 'vad', 'upload'] as const; export type RecordingMode = (typeof RECORDING_MODES)[number];

export const RECORDING_MODE_OPTIONS = [ { label: 'Manual', value: 'manual', icon: 'mic', desktopOnly: false }, { label: 'Voice Activated', value: 'vad', icon: 'mic-voice', desktopOnly: false, }, { label: 'Upload File', value: 'upload', icon: 'upload', desktopOnly: false }, ] as const satisfies { label: string; value: RecordingMode; icon: string; desktopOnly: boolean; }[];

// Derive IDs for validation if needed export const RECORDING_MODE_IDS = RECORDING_MODE_OPTIONS.map((o) => o.value);

Choosing a Pattern

Scenario Pattern

Label = formatted value (e.g., "128 kbps") Simple Values -> Derived

Label needs separate data (e.g., "16 kHz - Optimized for speech") Values + Metadata

Options have extra UI fields (icon, description, disabled) Rich Array

Platform-specific or runtime-conditional content Keep inline in component

Naming Rules

Source Arrays

• Use plural noun: PROVIDERS , MODES , LANGUAGES

• Add unit suffix when relevant: BITRATES_KBPS , SAMPLE_RATES

• Avoid redundant _VALUES suffix

Derived/Options Arrays

• Use plural noun + _OPTIONS suffix: BITRATE_OPTIONS , SAMPLE_RATE_OPTIONS

• For IDs: plural noun + _IDS suffix: PROVIDER_IDS

Label Maps

• Use singular _TO_LABEL suffix: LANGUAGES_TO_LABEL

• Describes the operation (id -> label), not the container

• Reads naturally: LANGUAGES_TO_LABEL[lang] = "get the label for this language"

Constant Casing

• Always use SCREAMING_SNAKE_CASE for exported constants

• Never use camelCase for constant objects/arrays

Co-location

Options arrays should be co-located with their source array in the same file. Avoid creating options inline in Svelte components; import pre-defined options instead.

Exception: Keep options inline when they have platform-specific or runtime-conditional content that would require importing platform constants into the data module.

Parameter Destructuring for Factory Functions

Prefer Parameter Destructuring Over Body Destructuring

When writing factory functions that take options objects, destructure directly in the function signature instead of in the function body. This is the established pattern in the codebase.

Bad Pattern (Body Destructuring)

// DON'T: Extra line of ceremony function createSomething(opts: { foo: string; bar?: number }) { const { foo, bar = 10 } = opts; // Unnecessary extra line return { foo, bar }; }

Good Pattern (Parameter Destructuring)

// DO: Destructure directly in parameters function createSomething({ foo, bar = 10 }: { foo: string; bar?: number }) { return { foo, bar }; }

Why This Matters

• Fewer lines: Removes the extra destructuring statement

• Defaults at API boundary: Users see defaults in the signature, not hidden in the body

• Works with const generics: TypeScript literal inference works correctly: function select<const TOptions extends readonly string[]>({ options, nullable = false, }: { options: TOptions; nullable?: boolean; }) { ... }

• Closures work identically: Inner functions capture the same variables either way

When Body Destructuring is Valid

• Need to distinguish "property missing" vs "property is undefined " ('key' in opts )

• Complex normalization/validation of the options object

• Need to pass the entire opts object to other functions

Codebase Examples

// From packages/epicenter/src/core/schema/columns.ts export function select<const TOptions extends readonly [string, ...string[]]>({ options, nullable = false, default: defaultValue, }: { options: TOptions; nullable?: boolean; default?: TOptions[number]; }): SelectColumnSchema<TOptions, boolean> { return { type: 'select', nullable, options, default: defaultValue }; }

// From apps/whispering/.../create-key-recorder.svelte.ts export function createKeyRecorder({ pressedKeys, onRegister, onClear, }: { pressedKeys: PressedKeys; onRegister: (keyCombination: KeyboardEventSupportedKey[]) => void; onClear: () => void; }) { ... }

Arktype Optional Properties

Never Use | undefined for Optional Properties

When defining optional properties in arktype schemas, always use the 'key?' syntax instead of | undefined unions. This is critical for JSON Schema conversion (used by OpenAPI/MCP).

Bad Pattern

// DON'T: Explicit undefined union - breaks JSON Schema conversion const schema = type({ window_id: 'string | undefined', url: 'string | undefined', });

This produces invalid JSON Schema with anyOf: [{type: "string"}, {}] because undefined has no JSON Schema equivalent.

Good Pattern

// DO: Optional property syntax - converts cleanly to JSON Schema const schema = type({ 'window_id?': 'string', 'url?': 'string', });

This correctly omits properties from the required array in JSON Schema.

Why This Matters

Syntax TypeScript Behavior JSON Schema

key: 'string | undefined'

Required prop, accepts string or undefined Broken (triggers fallback)

'key?': 'string'

Optional prop, accepts string Clean (omitted from required )

Both behave similarly in TypeScript, but only the ? syntax converts correctly to JSON Schema for OpenAPI documentation and MCP tool schemas.

Inline Definitions in Tests

Prefer Inlining Single-Use Definitions

When a schema, builder, or configuration is only used once in a test, inline it directly at the call site rather than extracting to a variable.

Bad Pattern (Extracted Variables)

test('creates workspace with tables', () => { const posts = defineTable(type({ id: 'string', title: 'string', _v: '1' }));

const theme = defineKv(type("'light' | 'dark'"), 'light');

const workspace = defineWorkspace({
	id: 'test-app',
	tables: { posts },
	kv: { theme },
});

expect(workspace.id).toBe('test-app');

});

Good Pattern (Inlined)

test('creates workspace with tables', () => { const workspace = defineWorkspace({ id: 'test-app', tables: { posts: defineTable(type({ id: 'string', title: 'string', _v: '1' })), }, kv: { theme: defineKv(type("'light' | 'dark'"), 'light'), }, });

expect(workspace.id).toBe('test-app');

});

Why Inlining is Better

• All context in one place: No scrolling to understand what posts or theme are

• Reduces naming overhead: No need to invent variable names for single-use values

• Matches mental model: The definition IS the usage - they're one conceptual unit

• Easier to copy/modify: Self-contained test setup is easier to duplicate and tweak

When to Extract

Extract to a variable when:

• The value is used multiple times in the same test

• You need to call methods on the result (e.g., posts.migrate() , posts.versions )

• The definition is shared across multiple tests in a beforeEach or test fixture

• The inline version would exceed ~15-20 lines and hurt readability

Applies To

• defineTable() , defineKv() , defineWorkspace() builders

• createTables() , createKV() factory calls

• Schema definitions (arktype, zod, etc.)

• Configuration objects passed to factories

• Mock functions used only once

Test File Organization

Shadow Source Files with Test Files

Each source file should have a corresponding test file in the same directory:

src/static/ ├── schema-union.ts ├── schema-union.test.ts # Tests for schema-union.ts ├── define-table.ts ├── define-table.test.ts # Tests for define-table.ts ├── create-tables.ts ├── create-tables.test.ts # Tests for create-tables.ts └── types.ts # No test file (pure types)

Benefits

• Clear ownership: Each test file tests exactly one source file

• Easy navigation: Find tests by looking next to the source

• Focused testing: Easier to run tests for just one module

• Maintainability: When source changes, you know which test file to update

What Gets Test Files

File Type Test File? Reason

Functions/classes with logic Yes Has behavior to test

Type definitions only No No runtime behavior

Re-export barrels (index.ts ) No Just re-exports, tested via consumers

Internal helpers Maybe Test via consumer if tightly coupled

Naming Convention

• Source: foo-bar.ts

• Test: foo-bar.test.ts

Integration Tests

For tests spanning multiple modules, either:

• Add to the test file of the highest-level consumer

• Create a dedicated [feature].integration.test.ts if substantial

Branded Types Pattern

Use Brand Constructors, Never Raw Type Assertions

When working with branded types (nominal typing), always create a brand constructor function. Never use as BrandedType assertions scattered throughout the codebase.

Bad Pattern (Scattered Assertions)

// types.ts type RowId = string & Brand<'RowId'>;

// file1.ts const id = someString as RowId; // Bad: assertion here

// file2.ts function getRow(id: string) { doSomething(id as RowId); // Bad: another assertion }

// file3.ts const parsed = key.split(':')[0] as RowId; // Bad: assertions everywhere

Good Pattern (Brand Constructor)

// types.ts import type { Brand } from 'wellcrafted/brand';

type RowId = string & Brand<'RowId'>;

// Brand constructor - THE ONLY place with as RowId // Uses PascalCase to match the type name (avoids parameter shadowing) function RowId(id: string): RowId { return id as RowId; }

// file1.ts const id = RowId(someString); // Good: uses constructor

// file2.ts function getRow(rowId: string) { doSomething(RowId(rowId)); // Good: no shadowing issues }

// file3.ts const parsed = RowId(key.split(':')[0]); // Good: consistent

Why Brand Constructors Are Better

• Single source of truth: Only one place has the type assertion

• Future validation: Easy to add runtime validation later

• Searchable: RowId( is easy to find and audit

• Explicit boundaries: Clear where unbranded -> branded conversion happens

• Refactor-safe: Change the branding logic in one place

• No shadowing: PascalCase constructor doesn't shadow camelCase parameters

Implementation Pattern

import type { Brand } from 'wellcrafted/brand';

// 1. Define the branded type export type RowId = string & Brand<'RowId'>;

// 2. Create the brand constructor (only as assertion in codebase) // PascalCase matches the type - TypeScript allows same-name type + value export function RowId(id: string): RowId { return id as RowId; }

// 3. Optionally add validation export function RowId(id: string): RowId { if (id.includes(':')) { throw new Error(RowId cannot contain ':': ${id}); } return id as RowId; }

Naming Convention

Branded Type Constructor Function

RowId

RowId()

FieldId

FieldId()

UserId

UserId()

DocumentGuid

DocumentGuid()

The constructor uses PascalCase matching the type name. TypeScript allows a type and value to share the same name (different namespaces). This avoids parameter shadowing issues.

When Functions Accept Branded Types

If a function requires a branded type, callers must use the brand constructor:

// Function requires branded RowId function getRow(id: RowId): Row { ... }

// Caller must brand the string - no shadowing since RowId() is PascalCase function processRow(rowId: string) { getRow(RowId(rowId)); // rowId param doesn't shadow RowId() function }

This makes type boundaries visible and intentional, without forcing awkward parameter renames.

Branded IDs for Workspace Tables

Every defineTable() schema MUST use branded ID types for the id field and all string foreign keys. Never use plain 'string' for table IDs.

For tables that use arktype schemas, follow the three-part pattern:

// 1. TYPE — extends Id, not string (enables single-cast in generator) export type SavedTabId = Id & Brand<'SavedTabId'>;

// 2. VALIDATOR — zero-cost type assertion for schema composition export const SavedTabId = type('string').as<SavedTabId>();

// 3. GENERATOR — wraps generateId() so the cast lives in one place export const generateSavedTabId = (): SavedTabId => generateId() as SavedTabId;

Use directly in the schema: id: SavedTabId and for optional FKs: 'parentId?': SavedTabId.or('undefined') .

At call sites, use the generator—never the double-cast:

// Good const id = generateSavedTabId();

// Bad — scattered casts const id = generateId() as string as SavedTabId;

The generate* prefix means "new ID from scratch." The create* prefix means "assemble from inputs" (e.g., createTabCompositeId(deviceId, tabId) ).

Not every branded type needs all three parts. DeviceId is set from an external source (no generator). Path types like AbsolutePath need only the type.

See the workspace-api skill for the full workspace file structure and rules.

Extract Coupled let State Into Sub-Factories

When a factory function accumulates let statements that are always read, written, and reset together, extract them into a sub-factory. The tell: two or three let declarations that move as a pack across multiple inner functions.

The Smell

function createProvider(config) { let retries = 0; let reconnectSleeper: Sleeper | null = null;

async function runLoop() { // ... 5-line backoff ceremony using retries + reconnectSleeper ... // ... appears in TWO places ... }

function handleOnline() { reconnectSleeper?.wake(); // reaches into closure state }

function handleSuccess() { retries = 0; // reset scattered across the function } }

retries and reconnectSleeper are one concept ("backoff") split across two let declarations, two inline ceremonies, and one external poke.

The Fix

Pull coupled state into its own factory with named methods:

function createBackoff() { let retries = 0; let sleeper: { promise: Promise<void>; wake(): void } | null = null;

return { async sleep() { /* compute delay, create sleeper, await, cleanup */ }, wake() { sleeper?.wake(); }, reset() { retries = 0; }, }; }

The parent factory replaces scattered let manipulation with named calls:

function createProvider(config) { const backoff = createBackoff();

async function runLoop() { await backoff.sleep(); // was 5 duplicated lines }

function handleOnline() { backoff.wake(); // was reconnectSleeper?.wake() }

function handleSuccess() { backoff.reset(); // was retries = 0 } }

When to Extract

Signal Action

Two+ let s always set in the same function Likely one concept

Resetting one requires resetting the others Definitely one concept

An external caller reaches into one of them The concept needs a public API

The same multi-line ceremony appears twice Extract and name it

When NOT to Extract

Don't extract let state that's deeply woven into control flow branching. If the variables are the loop's decision-making state (e.g., desired , runId , connectRun in a supervisor loop), extracting them just renames the complexity without reducing it. The test: do the call sites get simpler?

See docs/articles/let-packs-are-factories-waiting-to-be-named.md for a full walkthrough with three extractions from a real sync provider.

Const Generic Array Inference

Use const T extends readonly T[] to preserve literal types without requiring as const at call sites.

Pattern Plain ['a','b','c']

With as const

T extends string[]

string[]

["a", "b", "c"]

T extends readonly string[]

string[]

readonly ["a", "b", "c"]

const T extends string[]

["a", "b", "c"]

["a", "b", "c"]

const T extends readonly string[]

readonly ["a", "b", "c"]

readonly ["a", "b", "c"]

The const modifier preserves literal types; the readonly constraint determines mutability.

// From packages/epicenter/src/core/schema/fields/factories.ts export function select<const TOptions extends readonly [string, ...string[]]>({ id, options, }: { id: string; options: TOptions; }): SelectField<TOptions> { // ... }

// Caller gets literal union type — no as const needed const status = select({ id: 'status', options: ['draft', 'published'] }); // status.options[number] is "draft" | "published", not string

See docs/articles/typescript-const-modifier-generic-type-parameters.md for details.