typescript-dev
TypeScript Development
Type-safe code = compile-time errors = runtime confidence.
<when_to_use>
- Writing new TypeScript code
- Eliminating
anytypes - Using modern TypeScript 5.5+ features
- Validating API inputs/outputs with Zod
- Implementing Result types and discriminated unions
- Creating branded types for domain concepts
NOT for: runtime-only logic unrelated to types, non-TypeScript projects
</when_to_use>
tsconfig.json strict settings:
{
"compilerOptions": {
"strict": true,
"noUncheckedIndexedAccess": true,
"exactOptionalPropertyTypes": true,
"noImplicitOverride": true,
"noPropertyAccessFromIndexSignature": true,
"noFallthroughCasesInSwitch": true,
"noImplicitReturns": true,
"forceConsistentCasingInFileNames": true,
"verbatimModuleSyntax": true,
"isolatedModules": true,
"skipLibCheck": false
}
}
Version requirements: TS 5.2+ (using), TS 5.4+ (NoInfer), TS 5.5+ (inferred predicates)
Core Patterns
<eliminating_any>
any defeats the type system. Use unknown + guards.
// ❌ NEVER
function process(data: any) { return data.value; }
// ✅ ALWAYS
function process(data: unknown): string {
if (!hasValue(data)) throw new TypeError('Invalid');
return data.value.toString();
}
function hasValue(v: unknown): v is { value: unknown } {
return typeof v === 'object' && v !== null && 'value' in v;
}
Validate at boundaries:
async function fetchUser(id: string): Promise<User> {
const data: unknown = await fetch(`/api/users/${id}`).then(r => r.json());
return UserSchema.parse(data);
}
</eliminating_any>
<result_types>
Exceptions hide errors from types. Result makes them explicit.
type Result<T, E = Error> =
| { readonly ok: true; readonly value: T }
| { readonly ok: false; readonly error: E };
type UserError =
| { readonly type: 'not-found'; readonly id: string }
| { readonly type: 'network'; readonly message: string };
async function getUser(id: string): Promise<Result<User, UserError>> {
try {
const response = await fetch(`/api/users/${id}`);
if (response.status === 404)
return { ok: false, error: { type: 'not-found', id } };
if (!response.ok)
return { ok: false, error: { type: 'network', message: response.statusText } };
return { ok: true, value: await response.json() };
} catch (e) {
return { ok: false, error: { type: 'network', message: String(e) } };
}
}
// Caller must handle
const result = await getUser(id);
if (!result.ok) {
switch (result.error.type) {
case 'not-found': return showNotFound(result.error.id);
case 'network': return showError(result.error.message);
}
}
return renderUser(result.value);
See result-pattern.md for utilities (map, flatMap, combine).
</result_types>
<discriminated_unions>
Prevent illegal state combinations.
// ❌ Allows { status: 'loading', data: user, error: 'Failed' }
type Request = { status: 'idle'|'loading'|'success'|'error'; data?: User; error?: string; };
// ✅ Only valid states
type RequestState =
| { readonly status: 'idle' }
| { readonly status: 'loading' }
| { readonly status: 'success'; readonly data: User }
| { readonly status: 'error'; readonly error: string };
function render(state: RequestState): JSX.Element {
switch (state.status) {
case 'idle': return <div>Ready</div>;
case 'loading': return <div>Loading...</div>;
case 'success': return <div>{state.data.name}</div>;
case 'error': return <div>Error: {state.error}</div>;
default: return assertNever(state);
}
}
function assertNever(value: never): never {
throw new Error(`Unhandled: ${JSON.stringify(value)}`);
}
</discriminated_unions>
<branded_types>
Prevent mixing incompatible primitives.
declare const __brand: unique symbol;
type Brand<T, B extends string> = T & { readonly [__brand]: B };
type UserId = Brand<string, 'UserId'>;
type ProductId = Brand<string, 'ProductId'>;
function createUserId(value: string): UserId {
if (!/^user-\d+$/.test(value)) throw new TypeError(`Invalid: ${value}`);
return value as UserId;
}
const userId = createUserId('user-123');
// getUser(productId); // ❌ Type error
getUser(userId); // ✅ Works
Security:
type SanitizedHtml = Brand<string, 'SanitizedHtml'>;
function sanitize(raw: string): SanitizedHtml {
return escapeHtml(raw) as SanitizedHtml;
}
function render(html: SanitizedHtml): void {
element.innerHTML = html; // Type proves sanitization
}
See branded-types.md for advanced patterns.
</branded_types>
Modern TypeScript (5.2+)
<resource_management>
using for automatic cleanup (TS 5.2+):
class DatabaseConnection implements Disposable {
[Symbol.dispose]() { this.close(); }
}
function query() {
using conn = new DatabaseConnection();
return conn.query('SELECT * FROM users');
} // Automatically closed
async function asyncWork() {
await using resource = new AsyncResource();
} // Disposed with await
Use for: connections, file handles, locks, transactions.
</resource_management>
<satisfies_operator>
Validate type without widening (TS 4.9+):
const config = {
port: 3000,
host: 'localhost'
} satisfies Record<string, string | number>;
config.port // number (not string | number)
const routes = {
home: '/',
user: '/user/:id'
} as const satisfies Record<string, string>;
type HomeRoute = typeof routes.home; // '/'
</satisfies_operator>
<const_type_parameters>
Preserve literals through generics (TS 5.0+):
function makeTuple<const T extends readonly unknown[]>(...args: T): T {
return args;
}
const result = makeTuple('a', 'b', 'c'); // ['a', 'b', 'c'] not string[]
</const_type_parameters>
<inferred_predicates>
TS 5.5+ auto-infers type predicates:
function isString(x: unknown) {
return typeof x === 'string';
}
// Inferred: (x: unknown) => x is string
const strings = values.filter(isString); // string[]
</inferred_predicates>
<template_literals>
Pattern matching at type level:
type Route = `/${string}`;
type ApiRoute = `/api/v${number}/${string}`;
type ExtractParams<T extends string> =
T extends `${string}:${infer P}/${infer R}` ? P | ExtractParams<`/${R}`>
: T extends `${string}:${infer P}` ? P : never;
type Params = ExtractParams<'/user/:id/post/:postId'>; // 'id' | 'postId'
See modern-features.md for TS 5.5-5.8.
</template_literals>
Zod Validation
Schema = runtime validation + TypeScript type.
<zod_core>
import { z } from 'zod';
const UserSchema = z.object({
id: z.string().uuid(),
email: z.string().email(),
name: z.string().min(1).max(100)
});
type User = z.infer<typeof UserSchema>;
// safeParse preferred
const result = UserSchema.safeParse(data);
if (!result.success) {
console.error(result.error.issues);
return;
}
const user = result.data;
</zod_core>
<zod_patterns>
Discriminated unions (preferred over z.union):
const ApiResponse = z.discriminatedUnion("type", [
z.object({ type: z.literal("success"), data: z.unknown() }),
z.object({ type: z.literal("error"), code: z.string(), message: z.string() })
]);
Environment variables:
const EnvSchema = z.object({
NODE_ENV: z.enum(['development', 'production', 'test']).default('development'),
DATABASE_URL: z.string().url(),
PORT: z.coerce.number().int().positive().default(3000)
});
const env = EnvSchema.parse(process.env);
API validation (Hono):
import { zValidator } from '@hono/zod-validator';
app.post('/users', zValidator('json', UserSchema), (c) => {
const user = c.req.valid('json');
return c.json(user);
});
See:
- zod-building-blocks.md - primitives, refinements, transforms
- zod-schemas.md - composition patterns
- zod-integration.md - API/form/env integration
</zod_patterns>
Type Guards
// User-defined
function isString(v: unknown): v is string {
return typeof v === 'string';
}
// Assertion
function assertString(v: unknown): asserts v is string {
if (typeof v !== 'string') throw new TypeError('Expected string');
}
// With noUncheckedIndexedAccess
const users: User[] = getUsers();
const first = users[0]; // User | undefined
if (first !== undefined) processUser(first);
See advanced-types.md for utilities.
TSDoc
Types show structure. TSDoc shows intent. Critical for AI agents.
/**
* Authenticates user and returns session token.
* @param credentials - User login credentials
* @returns Session token valid for 24 hours
* @throws {AuthenticationError} Invalid credentials
* @example
* const token = await authenticate({ email, password });
*/
export async function authenticate(credentials: Credentials): Promise<SessionToken>;
Document: all exports, parameters with constraints, thrown errors, non-obvious returns.
See tsdoc-patterns.md for comprehensive guide.
ALWAYS:
- Strict TypeScript config enabled
- Type-only imports:
import type { User } from './types' - Const assertions for literal types
- Exhaustive matching with
assertNever - Runtime validation at boundaries (Zod)
- Branded types for domain/sensitive data
- Result types for error-prone operations
satisfiesfor literal inferenceusingfor resources with cleanup- TSDoc on all exports
NEVER:
any(useunknown+ guards)@ts-ignore(fix types or document)- TypeScript enums (use const assertions or z.enum)
- Non-null assertions
!(use guards) - Loose state (use discriminated unions)
- Hidden errors (use Result)
PREFER:
- safeParse over parse
- z.discriminatedUnion over z.union
- Inferred predicates (TS 5.5+)
- Const type parameters for literals
Type Patterns:
- result-pattern.md - Result/Either utilities
- branded-types.md - advanced branded patterns
- advanced-types.md - template literals, utilities
Modern Features:
- modern-features.md - TS 5.5-5.8
- migration-paths.md - upgrading TypeScript
Zod:
- zod-building-blocks.md - primitives, transforms
- zod-schemas.md - composition patterns
- zod-integration.md - API, forms, env
TSDoc:
- tsdoc-patterns.md - documentation patterns
Examples:
- api-response.md - end-to-end type-safe API
- form-validation.md - Zod + React Hook Form
- resource-management.md - using declarations
- state-machine.md - discriminated union patterns
More from outfitter-dev/agents
codebase-recon
This skill should be used when analyzing codebases, understanding architecture, or when "analyze", "investigate", "explore code", or "understand architecture" are mentioned.
93graphite-stacks
This skill should be used when the user asks to "create a stack", "submit stacked PRs", "gt submit", "gt create", "reorganize branches", "fix stack corruption", or mentions Graphite, stacked PRs, gt commands, or trunk-based development workflows.
76code-review
This skill should be used when reviewing code before commit, conducting quality gates, or when "review", "fresh eyes", "pre-commit review", or "quality gate" are mentioned.
34hono-dev
This skill should be used when building APIs with Hono, using hc client, implementing OpenAPI, or when "Hono", "RPC", or "type-safe API" are mentioned.
28software-craft
This skill should be used when making design decisions, evaluating trade-offs, assessing code quality, or when "engineering judgment" or "code quality" are mentioned.
28subagents
This skill should be used when coordinating agents, delegating tasks to specialists, or when "dispatch agents", "which agent", or "multi-agent" are mentioned.
25