sindresorhus/ky

Ky is a tiny and elegant HTTP client based on the Fetch API

Ky targets modern browsers, Node.js, Bun, and Deno.

It's just a tiny package with no dependencies.

• Simpler API
• Method shortcuts (ky.post())
• Treats non-2xx status codes as errors (after redirects)
• Retries failed requests
• JSON option
• Timeout support
• Upload and download progress
• Base URL option
• Instances with custom defaults
• Hooks
• Response validation with Standard Schema (Zod, Valibot, etc.)
• TypeScript niceties (e.g., .json() supports generics and defaults to unknown, not any)

npm install ky

• jsdelivr
• unpkg
• esm.sh

import ky from 'ky';

const json = await ky.post('https://example.com', {json: {foo: true}}).json();

console.log(json);
//=> {data: '🦄'}

With plain fetch, it would be:

class HTTPError extends Error {}

const response = await fetch('https://example.com', {
	method: 'POST',
	body: JSON.stringify({foo: true}),
	headers: {
		'content-type': 'application/json'
	}
});

if (!response.ok) {
	throw new HTTPError(`Fetch error: ${response.statusText}`);
}

const json = await response.json();

console.log(json);
//=> {data: '🦄'}

If you are using Deno, import Ky from a URL. For example, using a CDN:

import ky from 'https://esm.sh/ky';

The input and options are the same as fetch, with additional options available (see below).

Returns a Response object with Body methods added for convenience. So you can, for example, call ky.get(input).json() directly without having to await the Response first. When called like that, an appropriate Accept header will be set depending on the body method used. Unlike the Body methods of window.fetch, these will throw an HTTPError if the response status is not in the range of 200...299. Note that .json() will throw if the body is empty, as it cannot be parsed. However, you can use the parseJson option to add custom handling for empty bodies.

Available body shortcuts: .json(), .text(), .formData(), .arrayBuffer(), .blob(), and .bytes(). The .bytes() shortcut is only present when the runtime supports Response.prototype.bytes().

Invalid options, an invalid input, and an error thrown by an init hook reject the returned promise instead of throwing, the same way fetch() reports them. They are not passed to beforeError hooks, because no request was made.

import ky from 'ky';

const user = await ky('/api/user').json();

console.log(user);

⌨️ TypeScript: Accepts an optional type parameter, which defaults to unknown, and is passed through to the return type of .json().

import ky from 'ky';

// user1 is unknown
const user1 = await ky('/api/users/1').json();
// user2 is a User
const user2 = await ky<User>('/api/users/2').json();
// user3 is a User
const user3 = await ky('/api/users/3').json<User>();

console.log([user1, user2, user3]);

You can also get the response body as JSON and validate it with a Standard Schema compatible validator (for example, Zod 3.24+). This throws a SchemaValidationError when validation fails.

import ky, {SchemaValidationError} from 'ky';
import {z} from 'zod';

const userSchema = z.object({name: z.string()});

try {
	const user = await ky('/api/user').json(userSchema);
	console.log(user.name);
} catch (error) {
	if (error instanceof SchemaValidationError) {
		console.error(error.issues);
	}
}

// Get raw bytes (when supported by the runtime)
const bytes = await ky('/api/file').bytes();
console.log(bytes instanceof Uint8Array);

Sets options.method to the method name and makes a request.

⌨️ TypeScript: Accepts an optional type parameter for use with JSON responses (see ky()).

Type: string | URL | Request

Same as fetch input.

When using a Request instance as input, the prefix and baseUrl options are ignored. The searchParams option still applies.

Type: object

Same as fetch options, plus the following additional options:

Type: string
Default: 'get'

HTTP method used to make the request.

Internally, the standard methods (GET, POST, PUT, PATCH, HEAD, DELETE, and QUERY) are uppercased in order to avoid server errors due to case sensitivity.

Type: object and any other value accepted by JSON.stringify()

Shortcut for sending JSON. Use this instead of the body option. Accepts any plain object or value, which will be stringified using JSON.stringify() and sent in the body with the correct header set.

The Content-Type header is set to application/json unless you set a Content-Type in the headers option, which always takes precedence. A Content-Type on a Request input is replaced, because the Request constructor sets one automatically from its body.

Type: string | object<string, string | number | boolean | undefined> | Array<Array<string | number | boolean>> | URLSearchParams
Default: ''

Search parameters to include in the request URL. Setting this will merge with any existing search parameters in the input URL.

Accepts any value supported by URLSearchParams().

When passing an object, setting a value to undefined deletes the parameter, including from the input URL, even when a later option layer adds the parameter again. null values are preserved and converted to the string 'null'.

When input is a Request with a body, the body is sent as a stream, which requires request stream support and, in Chromium-based browsers, an HTTP/2 or HTTP/3 connection (streaming uploads over HTTP/1.1 fail with a network error, even over plain HTTP). The inherited body is dropped when the search parameters change the URL and the request has to be rebuilt in an environment that cannot reuse it, which is the case without request stream support, when keepalive is true, or when the effective mode is 'no-cors'. A search parameter value that leaves the URL unchanged does not rebuild the request, so the body is kept. A compatible body passed explicitly with the body option is still used.

Type: string | URL

A base URL to resolve the input against. When the input (after applying the prefix option) is only a relative URL, such as 'users', '/users', or '//my-site.com', it will be resolved against the baseUrl to determine the destination of the request. Otherwise, the input is absolute, such as 'https://my-site.com', and it will bypass the baseUrl.

Useful when used with ky.extend() to create niche-specific Ky instances.

If the baseUrl itself is relative, it will be resolved against the environment's base URL, such as document.baseURI in browsers or location.href in Deno (see the --location flag).

Tip: When setting a baseUrl that has a path, we recommend that it include a trailing slash /, as in '/api/' rather than /api. This ensures more intuitive behavior for page-relative input URLs, such as 'users' or './users', where they will extend from the full path of the baseUrl rather than replacing its last path segment.

import ky from 'ky';

// On https://example.com

const response = await ky('users', {baseUrl: '/api/'});
//=> 'https://example.com/api/users'

const response = await ky('/users', {baseUrl: '/api/'});
//=> 'https://example.com/users'

See FAQ: baseUrl vs prefix

Type: string | URL

A prefix to prepend to the input before making the request (and before it is resolved against the baseUrl). It can be any valid path or URL, either relative or absolute. A trailing slash / is optional and will be added automatically, if needed, when it is joined with input. Only takes effect when input is a string.

Useful when used with ky.extend() to create niche-specific Ky instances.

In most cases, you should use the baseUrl option instead, as it is more consistent with web standards. However, prefix is useful if you want origin-relative input URLs, such as /users, to be treated as if they were page-relative. In other words, the leading slash of the input will essentially be ignored, because the prefix will become part of the input before URL resolution happens.

See FAQ: baseUrl vs prefix

import ky from 'ky';

// On https://example.com

const response = await ky('users', {prefix: '/api/'});
//=> 'https://example.com/api/users'

const response = await ky('/users', {prefix: '/api/'});
//=> 'https://example.com/api/users'

Notes:

• The prefix and input are joined with a slash /, and slashes are normalized at the join boundary by trimming trailing slashes from prefix and leading slashes from input.
• After prefix and input are joined, the result is resolved against the baseUrl option, if present.

Type: object | number
Default:

• limit: 2
• methods: get put head delete options trace query
• statusCodes: 408 413 429 500 502 503 504
• afterStatusCodes: 413, 429, 503
• maxRetryAfter: Infinity
• backoffLimit: Infinity
• delay: attemptCount => 0.3 * (2 ** (attemptCount - 1)) * 1000
• jitter: undefined
• retryOnTimeout: false
• shouldRetry: undefined

Controls retry behavior. Each field is documented individually below.

If retry is a number, it will be used as limit and other defaults will remain in place.

retry.limit, including numeric shorthand, must be a finite, non-negative integer.

Network errors (e.g., DNS failures, connection refused, offline) are automatically retried for retriable methods. Only errors recognized as network errors are retried; other errors (e.g., programming bugs) are thrown immediately. Use shouldRetry to customize this behavior.

A hook can change retry.limit through options. Ky reads the limit again when each attempt starts. A beforeRequest or beforeRetry hook runs before the attempt starts, so it can lower or raise the limit. An afterResponse hook runs after the attempt started, so it can lower the limit, for example to stop retrying once a response says so. A higher limit set there only applies from the next attempt.

413 Payload Too Large is only retried when the response includes a retry timing header, unless shouldRetry returns true.

When the response status is contained in afterStatusCodes and the retry is allowed by statusCodes or shouldRetry, Ky uses retry timing headers to choose the retry delay. Retry-After may provide a delay in seconds or an HTTP-date. If Retry-After is missing, Ky falls back to rate-limit timing headers (RateLimit-Reset, X-RateLimit-Retry-After, X-RateLimit-Reset, and X-Rate-Limit-Reset). Numeric Retry-After and X-RateLimit-Retry-After values are interpreted as delay seconds. Numeric RateLimit-Reset, X-RateLimit-Reset, and X-Rate-Limit-Reset values may also be interpreted as Unix timestamps, from 2001-09-09 onwards. A value below that is read as delay seconds, and a timestamp already in the past means the retry happens immediately. If the status code is not in afterStatusCodes, retry timing headers will be ignored.

If the retry delay from a retry timing header is greater than maxRetryAfter, Ky will use maxRetryAfter.

The backoffLimit option is the upper limit of the delay per retry in milliseconds, for the delay calculated by retry.delay. To clamp the delay, set backoffLimit to 1000, for example. A delay from a retry timing header is not subject to this limit; use maxRetryAfter to bound it. By default, the delay is calculated with 0.3 * (2 ** (attemptCount - 1)) * 1000. The delay increases exponentially.

The delay option can be used to change how the delay between retries is calculated. The function receives one parameter, the attempt count, starting at 1, and must return the delay in milliseconds. A value that is not a non-negative number or Infinity throws a TypeError rather than collapsing to a 1ms retry, and the error that was being retried is kept as its cause so beforeError can still reach it.

The jitter option adds random jitter to retry delays to prevent thundering herd problems. When many clients retry simultaneously (e.g., after hitting a rate limit), they can overwhelm the server again. Jitter adds randomness to break this synchronization. Set to true to use full jitter, which randomizes the delay between 0 and the computed delay. Alternatively, pass a function to implement custom jitter strategies.

Note: Jitter is not applied when the server provides a retry timing header, as the server's explicit timing should be respected.

The retryOnTimeout option determines whether to retry when a request times out before a response is returned. By default, retries are not triggered following a timeout. Timeouts while reading a response body through Ky shortcut methods are not retried because the response has already been received.

The shouldRetry option provides custom retry logic that takes precedence over the default retry checks (retryOnTimeout, status code checks, etc.) for retriable methods. It is only called after the retry limit and method checks pass.

Note: This is different from the beforeRetry hook:

• shouldRetry: Controls WHETHER to retry (called before the retry decision is made)
• beforeRetry: Called AFTER retry is confirmed, allowing you to modify the request

The function receives a state object with the error and retry count (starts at 1 for the first retry), and should return:

• true to force a retry (bypasses retryOnTimeout, status code checks, and other default validations)
• false to prevent a retry (no retry will occur)
• undefined (or nothing) to use the default retry logic (retryOnTimeout, status codes, network errors). Unrecognized error types are not retried.

Returning true overrides retry eligibility, not timing. For HTTP errors with a status in afterStatusCodes, retry timing headers and maxRetryAfter still apply, without jitter. Otherwise, Ky uses the configured backoff delay.

General example

import ky from 'ky';

const json = await ky('https://example.com', {
	retry: {
		limit: 10,
		methods: ['get'],
		statusCodes: [500],
		backoffLimit: 3000
	}
}).json();

Retrying on timeout:

import ky from 'ky';

const json = await ky('https://example.com', {
	timeout: 5000,
	retry: {
		limit: 3,
		retryOnTimeout: true
	}
}).json();

Using jitter to prevent thundering herd:

import ky from 'ky';

const json = await ky('https://example.com', {
	retry: {
		limit: 5,

		// Full jitter (randomizes delay between 0 and computed value)
		jitter: true

		// Percentage jitter (80-120% of delay)
		// jitter: delay => delay * (0.8 + Math.random() * 0.4)

		// Absolute jitter (±100ms)
		// jitter: delay => delay + (Math.random() * 200 - 100)
	}
}).json();

Custom retry logic:

import ky, {HTTPError} from 'ky';

const json = await ky('https://example.com', {
	retry: {
		limit: 3,
		shouldRetry: ({error, retryCount}) => {
			// Retry on specific business logic errors from API
			if (error instanceof HTTPError) {
				const status = error.response.status;

				// Retry on 429 (rate limit) but only for first 2 attempts
				if (status === 429 && retryCount <= 2) {
					return true;
				}

				// Don't retry on 4xx errors except rate limits
				if (status >= 400 && status < 500) {
					return false;
				}
			}

			// Use default retry logic for other errors
			return undefined;
		}
	}
}).json();

Type: number | false
Default: 10000

Per-attempt timeout in milliseconds for getting a response, applied independently to each retry. Ky shortcut methods also use this value as a separate timeout for reading the response body. Must be a non-negative number or false, or a TypeError is thrown. A value greater than 2147483647, Infinity included, throws a RangeError. See also totalTimeout.

If set to false, there will be no per-attempt timeout.

If the signal you passed in is aborted while a body method is reading, or just after the read finished, the body method rejects with the abort reason rather than resolving with the bytes that already arrived.

Type: number | false
Default: false

Overall timeout in milliseconds for the entire operation, including retries and delays. Throws a TimeoutError if exceeded. Must be a non-negative number or false, or a TypeError is thrown. A value greater than 2147483647, Infinity included, throws a RangeError.

beforeError hooks run after an error is produced and are not bounded by totalTimeout.

If set to false or not specified, there is no overall timeout.

import ky from 'ky';

// Each attempt gets 5s, but the whole operation must complete within 30s
const json = await ky('https://example.com', {
	timeout: 5000,
	totalTimeout: 30_000,
	retry: {
		limit: 3,
		retryOnTimeout: true,
	}
}).json();

Type: number
Default: Infinity

Maximum response body size in bytes. Must be a non-negative safe integer or Infinity. Set to 0 to allow only empty bodies.

The limit counts bytes from the response stream after decompression, independently of Content-Length. It applies as the body is consumed, including in afterResponse hooks and for responses returned by hooks. Exceeding the limit cancels the stream and throws a ResponseSizeError, without automatically retrying.

With await ky(url), the response can resolve before the limit is exceeded; the body read will reject instead. This limits body bytes, not total memory usage. Parsing, buffering, and concurrent requests can use additional memory.

Responses with a null body status, such as 204 and 205, are never wrapped, because the Response constructor rejects a body for those statuses. A runtime that still exposes a body for such a status, such as WebKit for 205, is therefore not limited.

import ky from 'ky';

const data = await ky('https://example.com/data', {
	maxResponseSize: 20 * 1024 * 1024,
}).json();

An undici Agent or Dispatcher, forwarded to fetch() as-is. Ky does not use it, it just passes it through.

A framework-specific fetch extension, such as Next.js's next, forwarded to fetch() as-is.

Ky forwards every option it does not recognize to fetch(). The type of such an option comes from the global RequestInit type, which @types/node extends with dispatcher and Next.js extends with next. An extension that your environment does not declare needs a type assertion.

Type: object<string, Function[]>
Default: {init: [], beforeRequest: [], beforeRetry: [], beforeError: [], afterResponse: []}

Hooks allow modifications during the request lifecycle. Hook functions may be async and are run serially, unless otherwise noted.

Each hook must be an array of functions. A single function, a string, or any other value, null included, throws a TypeError rather than being silently dropped. undefined means absent, so it clears the hooks it would have replaced.

Type: Function[]
Default: []

This hook enables you to modify the options before they are used to construct the request. The hook function receives the mutable options object and can modify it in place. You could, for example, modify searchParams, headers, or json here. The headers option starts as a plain object with lowercase names, where a header removed with undefined keeps an undefined value.

Ky gives the hook its own copies of headers, context, json, searchParams and retry, so changing them in place only affects the current request. The context copy is shallow, the same as in every other hook, so nested context values, such as a cache, stay shared across requests. The body option is not copied, so mutating a FormData or URLSearchParams body in place changes your own object and repeats on every request from the same instance. Assign a new instance instead, as shown in Modifying FormData in hooks. A value that a hook assigns is used as is, so assign a new object rather than one that you change later.

Unlike other hooks, init hooks are synchronous. An error thrown by one rejects the returned promise and is not passed to beforeError hooks.

A common use case is to add a search parameter to every request:

import ky from 'ky';

const api = ky.extend({
	hooks: {
		init: [
			options => {
				options.searchParams = {apiKey: getApiKey()};
			},
		],
	},
});

const response = await api.get('https://example.com/api/users');
// URL: https://example.com/api/users?apiKey=123

Type: Function[]
Default: []

This hook enables you to modify the request right before it is sent. Ky will make no further changes to the request after this. The hook function receives a state object with the normalized request, options, and retry count. You could, for example, modify request.headers here.

The retryCount is always 0, since beforeRequest hooks run once before retry handling begins.

The hook can return a Request to replace the outgoing request (remaining hooks will still run with the updated request). It can also return a Response to completely avoid making an HTTP request, in which case remaining beforeRequest hooks are skipped. This can be used to mock a request, check an internal cache, etc. A returned Request instance's signal is replaced with Ky's managed signal so that timeouts and user-provided abort signals still work.

Any error thrown by beforeRequest hooks is treated as fatal and will not trigger Ky's retry logic.

import ky from 'ky';

const api = ky.extend({
	hooks: {
		beforeRequest: [
			({request}) => {
				request.headers.set('Authorization', 'token initial-token');
			}
		]
	}
});

const response = await api.get('https://example.com/api/users');

Modifying the request URL:

import ky from 'ky';

const api = ky.extend({
	hooks: {
		beforeRequest: [
			({request}) => {
				const url = new URL(request.url);
				url.searchParams.set('token', 'secret-token');
				return new Request(url, request);
			}
		]
	}
});

const response = await api.get('https://example.com/api/users');

Type: Function[]
Default: []

This hook enables you to modify the request right before retry. Ky will make no further changes to the request after this. The hook function receives a state object with the normalized request, options, an error instance, and retry count. You could, for example, modify request.headers here.

The hook can return a Request to replace the outgoing retry request, or return a Response to skip the retry and use that response instead. A returned Request instance's signal is replaced with Ky's managed signal so that timeouts and user-provided abort signals still work. Note: Returning a request or response skips remaining beforeRetry hooks.

The retryCount is always >= 1, since this hook is only called during retries, not on the initial request.

If the request received a response, the error will be of type HTTPError. The Response object will be available at error.response, and the pre-parsed response body may be available at error.data. Be aware that some types of errors, such as network errors, inherently mean that a response was not received. In that case, the error will be an instance of NetworkError instead of HTTPError.

You can prevent Ky from retrying the request by throwing an error. Ky will not handle it in any way and the error will be propagated to the request initiator. The rest of the beforeRetry hooks will not be called in this case. Alternatively, you can return the ky.stop symbol to do the same thing but without propagating an error (this has some limitations, see ky.stop docs for details).

Changing options:

Return {options: {...}} to change onUploadProgress, onDownloadProgress, timeout, fetch, or throwHttpErrors for this retry and the ones after it. Omitted options keep their value, and undefined restores Ky's default, not the value from ky.create() or ky.extend(). Any other option, or an invalid value, throws an error. The remaining beforeRetry hooks still run and see the new progress callbacks in options. The options a hook receives never include timeout, fetch, or throwHttpErrors.

For example, turn off upload progress after a network error. Upload progress needs a streamed request body, which Chromium only sends over HTTP/2 or HTTP/3, so the upload fails on an HTTP/1.1 connection. Only retry an upload when it is safe to send it again, since a network error does not tell whether the server got the request.

import ky, {isNetworkError} from 'ky';

await ky.put(uploadUrl, {
	body: file,
	onUploadProgress: updateProgress,
	retry: {limit: 1},
	hooks: {
		beforeRetry: [
			({error}) => {
				if (isNetworkError(error)) {
					return {options: {onUploadProgress: undefined}};
				}
			}
		]
	}
});

Modifying headers:

import ky from 'ky';

const response = await ky('https://example.com', {
	hooks: {
		beforeRetry: [
			async ({request, options, error, retryCount}) => {
				const token = await ky('https://example.com/refresh-token');
				request.headers.set('Authorization', `token ${token}`);
			}
		]
	}
});

Modifying the request URL:

import ky, {isHTTPError} from 'ky';

const response = await ky('https://example.com/api', {
	hooks: {
		beforeRetry: [
			({request, error}) => {
				// Add query parameters based on error response
				if (
					isHTTPError(error)
					&& typeof error.data === 'object'
					&& error.data !== null
					&& 'processId' in error.data
				) {
					const url = new URL(request.url);
					url.searchParams.set('processId', String(error.data.processId));
					return new Request(url, request);
				}
			}
		]
	}
});

Returning a cached response:

import ky from 'ky';

const response = await ky('https://example.com/api', {
	hooks: {
		beforeRetry: [
			({error, retryCount}) => {
				// Use cached response instead of retrying
				if (retryCount > 1 && cachedResponse) {
					return cachedResponse;
				}
			}
		]
	}
});

Type: Function[]
Default: []

This hook enables you to modify any error right before it is thrown. The hook function receives a state object with the current request, the normalized Ky options, the error, and retry count, and should return an Error instance.

This hook receives errors Ky handles during its request lifecycle, including HTTPError, NetworkError, TimeoutError, ResponseSizeError, and ForceRetryError (when retry limit is exceeded via ky.retry()). Use type guards like isHTTPError(), isNetworkError(), isTimeoutError(), or isResponseSizeError() to handle specific error types.

Errors raised while consuming an already returned Response are outside the request lifecycle. Use a body method shortcut like await ky(url).json() to process body-read ResponseSizeError, NetworkError, and TimeoutError instances through this hook.

The retryCount is 0 for the initial request and increments with each retry. This allows you to distinguish between the initial request and retries, which is useful when you need different error handling based on retry attempts (e.g., showing different error messages on the final attempt).

If a beforeRequest or beforeRetry hook returns a new Request, inspect request for the final request state. options remains Ky's normalized options and may not mirror every property of a replacement Request.

import ky, {isHTTPError} from 'ky';

await ky('https://example.com', {
	hooks: {
		beforeError: [
			({request, options, error}) => {
				if (isHTTPError(error)) {
					if (
						typeof error.data === 'object'
						&& error.data !== null
						&& 'message' in error.data
					) {
						error.name = 'GitHubError';
						error.message = `${String(error.data.message)} (${error.response.status})`;
					}
				}

				// `request` and `options` are always available
				console.log(`Request to ${request.url} failed`, options.context);

				return error;
			}
		]
	}
});

Type: Function[]
Default: []

This hook enables you to read and optionally modify the response. The hook function receives a state object with the normalized request, options, a clone of the response, and retry count. The return value of the hook function will be used by Ky as the response object if it's an instance of Response.

You can also force a retry by returning ky.retry(options). This is useful when you need to retry based on the response body content, even if the response has a successful status code. The retry will respect the retry.limit option and be observable in beforeRetry hooks.

Any non-ky.retry() error thrown by afterResponse hooks is treated as fatal and will not trigger Ky's retry logic.

The retryCount is 0 for the initial request and increments with each retry. This allows you to distinguish between the initial request and retries, which is useful when you need different behavior for retries (e.g., showing a notification only on the final retry).

import ky from 'ky';

const response = await ky('https://example.com', {
	hooks: {
		afterResponse: [
			({response}) => {
				// You could do something with the response, for example, logging.
				log(response);

				// Or return a `Response` instance to overwrite the response.
				return new Response('A different response', {status: 200});
			},

			// Or retry with a fresh token on a 401 error
			async ({request, response, retryCount}) => {
				if (response.status === 401 && retryCount === 0) {
					// Only refresh on first 401, not on subsequent retries
					const {token} = await ky.post('https://example.com/auth/refresh').json();

					const headers = new Headers(request.headers);
					headers.set('Authorization', `Bearer ${token}`);

					return ky.retry({
						request: new Request(request, {headers}),
						code: 'TOKEN_REFRESHED'
					});
				}
			},

			// Or force retry based on response body content
			async ({response}) => {
				if (response.status === 200) {
					const data = await response.json();
					if (data.error?.code === 'RATE_LIMIT') {
						// Retry with custom delay from API response
						return ky.retry({
							delay: data.error.retryAfter * 1000,
							code: 'RATE_LIMIT'
						});
					}
				}
			},

			// Or show a notification only on the last retry for 5xx errors
			({options, response, retryCount}) => {
				if (response.status >= 500 && response.status <= 599) {
					if (retryCount === options.retry.limit) {
						showNotification('Request failed after all retries');
					}
				}
			}
		]
	}
});

Type: boolean | (status: number) => boolean
Default: true

Throw an HTTPError when, after following redirects, the response has a non-2xx status code. To also throw for redirects instead of following them, set the redirect option to 'manual'.

Setting this to false may be useful if you are checking for resource availability and are expecting error responses.

You can also pass a function that accepts the HTTP status code and returns a boolean for selective error handling. Note that this can violate the principle of least surprise, so it's recommended to use the boolean form unless you have a specific use case like treating 404 responses differently.

Note: If false, error responses are considered successful and the request will not be retried.

Note: Opaque responses from no-cors requests are returned as-is (without throwing HTTPError), since the actual status is hidden by the browser.

Type: Function

Download progress event handler.

The function receives these arguments:

• progress is an object with these properties:
  • percent is a number between 0 and 1 representing the progress percentage.
  • transferredBytes is the number of bytes transferred so far.
  • totalBytes is the total number of bytes to be transferred. This is an estimate and may be 0 for an empty transfer or when the total size cannot be determined.
• chunk is an instance of Uint8Array containing the data that was received. When an empty response body stream completes, the callback receives an empty chunk.

content-length is only used as the total for a response that is not content-coded, since it counts encoded bytes while the progress stream counts the bytes after decompression. A compressed response therefore reports totalBytes: 0 until it completes, so the percentage cannot be calculated while downloading. For a cross-origin response in a browser, content-encoding is only visible when the server lists it in Access-Control-Expose-Headers. Otherwise Ky cannot tell that the response is compressed, so it uses content-length and the percentage stays close to 100% for most of the download.

Responses with no body at all are not streamed, so no progress events are emitted for them. That covers a null body status such as 204, and a HEAD response in runtimes that give it no body, such as browsers, Node.js and Deno. Bun gives a HEAD response an empty body, so it reports one final event with transferredBytes: 0. A response whose body an afterResponse hook already read, or locked with a reader, is passed through unchanged, so it reports no progress either.

Do not throw from the callback. A throw fails the response stream, which the runtime reports in its own way: Node.js rejects the body read with your error, while Chromium reports a network failure, so Ky's body methods throw a NetworkError there. Pass a signal from an AbortController to cancel instead.

import ky from 'ky';

const response = await ky('https://example.com', {
	onDownloadProgress: (progress, chunk) => {
		// Example output:
		// `100% - 1271 of 1271 bytes`
		console.log(`${progress.percent * 100}% - ${progress.transferredBytes} of ${progress.totalBytes} bytes`);
	}
});

Type: Function

Upload progress event handler.

The function receives these arguments:

• progress is an object with these properties:
  • percent is a number between 0 and 1 representing the progress percentage.
  • transferredBytes is the number of bytes transferred so far.
  • totalBytes is the total number of bytes to be transferred. This is an estimate and may be 0 for an empty transfer or when the total size cannot be determined.
• chunk is an instance of Uint8Array containing the data that was sent. When an empty request body stream completes, the callback receives an empty chunk.

A ReadableStream body cannot be measured, so totalBytes falls back to a content-length header you set. Browsers do not allow that header on a request, so there the total stays 0. When a hook replaces the request, totalBytes is still estimated from the body option until the final event, which reports the real total.

Do not throw from the callback. A throw fails the upload stream, which the runtime reports as a network failure, so Ky throws a NetworkError (in Node.js, with your error in its cause chain) and retries it like any other network error, running the callback again. Pass a signal from an AbortController to cancel instead.

import ky from 'ky';

const response = await ky.post('https://example.com/upload', {
	body: largeFile,
	onUploadProgress: (progress, chunk) => {
		// Example output:
		// `100% - 1271 of 1271 bytes`
		console.log(`${progress.percent * 100}% - ${progress.transferredBytes} of ${progress.totalBytes} bytes`);
	}
});

Type: Function
Default: JSON.parse()

User-defined JSON-parsing function.

The function receives the response text as the first argument and a context object as the second argument containing the request (Request) and response (Response).

Use-cases:

1. Parse JSON via the bourne package to protect from prototype pollution.
2. Parse JSON with reviver option of JSON.parse().
3. Log or handle JSON parse errors with request context.

import ky from 'ky';
import bourne from '@hapijs/bourne';

const json = await ky('https://example.com', {
	parseJson: text => bourne(text)
}).json();

import ky from 'ky';

const json = await ky('https://example.com', {
	parseJson: (text, {request, response}) => {
		console.log(`Parsing JSON from ${request.url} (status: ${response.status})`);
		return JSON.parse(text);
	}
}).json();

Type: Function
Default: JSON.stringify()

User-defined JSON-stringifying function.

Use-cases:

1. Stringify JSON with a custom replacer function.

import ky from 'ky';
import {DateTime} from 'luxon';

const json = await ky('https://example.com', {
	stringifyJson: data => JSON.stringify(data, (key, value) => {
		if (key.endsWith('_at')) {
			return DateTime.fromISO(value).toSeconds();
		}

		return value;
	})
}).json();

Type: Function
Default: fetch

User-defined fetch function. Has to be fully compatible with the Fetch API standard. It must resolve with a Response, or a TypeError is thrown.

Use-cases:

1. Use the fetch wrapper function provided by some frameworks that use server-side rendering (SSR).
2. Add custom instrumentation or logging to all requests.

import ky from 'ky';

const api = ky.create({
	fetch: async (request, init) => {
		const start = performance.now();
		const response = await fetch(request, init);
		const duration = performance.now() - start;
		console.log(`${request.method} ${request.url} - ${response.status} (${Math.round(duration)}ms)`);
		return response;
	}
});

const json = await api('https://example.com').json();

Type: object<string, unknown>
Default: {}

User-defined data passed to hooks.

This option allows you to pass arbitrary contextual data to hooks without polluting the request itself. The context is available in all hooks and is guaranteed to always be an object (never undefined), so you can safely access properties without optional chaining.

Use cases:

• Pass authentication tokens or API keys to hooks
• Attach request metadata for logging or debugging
• Implement conditional logic in hooks based on the request context
• Pass serverless environment bindings (e.g., Cloudflare Workers)

Note: Context is shallow merged. Top-level properties are merged, but nested objects are replaced. Only enumerable properties are copied.

import ky from 'ky';

// Pass data to hooks
const api = ky.create({
	hooks: {
		beforeRequest: [
			({request, options}) => {
				const {token} = options.context;
				if (token) {
					request.headers.set('Authorization', `Bearer ${token}`);
				}
			}
		]
	}
});

await api('https://example.com', {
	context: {
		token: 'secret123'
	}
}).json();

// Shallow merge: only top-level properties are merged
const instance = ky.create({
	context: {
		a: 1,
		b: {
			nested: true
		}
	}
});

const extended = instance.extend({
	context: {
		b: {
			updated: true
		},
		c: 3
	}
});
// Result: {a: 1, b: {updated: true}, c: 3}
// Note: The original `b.nested` is gone (shallow merge)

Create a new ky instance with some defaults overridden with your own.

In contrast to ky.create(), ky.extend() inherits defaults from its parent.

You can pass headers as a Headers instance or a plain object. Headers are normalized to a plain object with lowercase names when options are merged, so init hooks start with a plain object with lowercase keys. A header removed with undefined stays in that object with an undefined value.

You can remove a header with .extend() by passing the header with an undefined value. Passing undefined as a string removes the header only if it comes from a Headers instance.

Similarly, you can remove existing hooks entries by extending the hook with an explicit undefined.

import ky from 'ky';

const url = 'https://sindresorhus.com';

const original = ky.create({
	headers: {
		rainbow: 'rainbow',
		unicorn: 'unicorn'
	},
	hooks: {
		beforeRequest: [ () => console.log('before 1') ],
		afterResponse: [ () => console.log('after 1') ],
	},
});

const extended = original.extend({
	headers: {
		rainbow: undefined
	},
	hooks: {
		beforeRequest: undefined,
		afterResponse: [
			({request}) => {
				console.log('after 2');
				console.log(request.headers.has('rainbow'));
				console.log(request.headers.has('unicorn'));
			}
		],
	}
});

await extended(url);
//=> after 1
//=> after 2
//=> false
//=> true

You can also refer to parent defaults by providing a function to .extend().

import ky from 'ky';

const api = ky.create({prefix: 'https://example.com/api'});

const usersApi = api.extend((options) => ({prefix: `${options.prefix}/users`}));

const response = await usersApi.get('123');
//=> 'https://example.com/api/users/123'

const response = await api.get('version');
//=> 'https://example.com/api/version'

By default, .extend() deep-merges options: hooks are appended, headers are merged, and search parameters are accumulated. Use replaceOption when you want to fully replace a merged property instead.

When extending an instance, signals are combined. Use replaceOption(signal) to replace inherited signals, or signal: undefined to remove them. null is accepted for RequestInit compatibility and is treated like an absent signal, so it does not remove an inherited signal.

import ky, {replaceOption} from 'ky';

const api = ky.create({
	hooks: {
		beforeRequest: [addAuth, addTracking],
	},
});

// Appends as expected
const extended = api.extend({hooks: {beforeRequest: [logRequest]}});
// extended hooks.beforeRequest is [addAuth, addTracking, logRequest]

// Replaces instead of appending
const replaced = api.extend({hooks: replaceOption({beforeRequest: [onlyThis]})});
// replaced hooks.beforeRequest is [onlyThis]

Create a new Ky instance with complete new defaults, without inheriting from any parent instance.

import ky from 'ky';

// On https://my-site.com

const api = ky.create({baseUrl: 'https://example.com/api/'});

const response = await api.get('users/123');
//=> 'https://example.com/api/users/123'

const response = await api.get('status', {baseUrl: ''});
//=> 'https://my-site.com/status'

Type: object

A Symbol that can be returned by a beforeRetry hook to stop the retry. This will also short circuit the remaining beforeRetry hooks.

Note: Returning this symbol makes Ky abort and return with an undefined response. Be sure to check for a response before accessing any properties on it or use optional chaining. It is also incompatible with body methods, such as .json() or .text(), because there is no response to parse. Calling one throws a TypeError explaining this. In general, we recommend throwing an error instead of returning this symbol, as that will cause Ky to abort and then throw, which avoids these limitations.

A valid use-case for ky.stop is to prevent retries when making requests for side effects, where the returned data is not important. For example, logging client activity to the server.

import ky from 'ky';

// Note that `response` will be `undefined` in case `ky.stop` is returned.
const response = await ky.post('https://example.com', {
	hooks: {
		beforeRetry: [
			async ({request, options, error, retryCount}) => {
				const shouldStopRetry = await ky('https://example.com/api');
				if (shouldStopRetry) {
					return ky.stop;
				}
			}
		]
	}
});

// Using `.text()` or other body methods is not supported, because there is no response to parse.

Force a retry from an afterResponse hook.

This allows you to retry a request based on the response content, even if the response has a successful status code. The retry will respect the retry.limit option and skip the shouldRetry check. The forced retry is observable in beforeRetry hooks, where the error will be a ForceRetryError.

A marker returned by a second copy of Ky, which happens with a duplicated dependency, is still recognised, the same way the Ky error guards accept branded errors.

Type: object

Type: number

Custom delay in milliseconds before retrying. If not provided, uses the default retry delay calculation based on retry.delay configuration.

Note: Custom delays bypass jitter and backoffLimit. This is intentional, as custom delays often come from server responses (e.g., Retry-After headers) and should be respected exactly as specified.

Must be a non-negative number, Infinity included, or undefined; anything else makes the request throw a TypeError rather than collapsing to a 1ms retry.

Type: string

Error code for the retry.

This machine-readable identifier will be included in the error message passed to beforeRetry hooks, allowing you to distinguish between different types of forced retries. Anything other than a string makes the request throw a TypeError.

return ky.retry({code: 'RATE_LIMIT'});
// Resulting error message: 'Forced retry: RATE_LIMIT'

Type: Error

Original error that caused the retry. This allows you to preserve the error chain when forcing a retry based on caught exceptions. The error will be set as the cause of the ForceRetryError, enabling proper error chain traversal.

try {
	const data = await response.json();
	validateBusinessLogic(data);
} catch (error) {
	return ky.retry({
		code: 'VALIDATION_FAILED',
		cause: error  // Preserves original error in chain
	});
}

Type: Request

Custom request to use for the retry.

This allows you to modify or completely replace the request during a forced retry. The custom request becomes the starting point for the retry - beforeRetry hooks can still further modify it if needed. Anything other than a Request or undefined makes the request throw a TypeError.

Note: The custom request's signal will be replaced with Ky's managed signal to handle timeouts and user-provided abort signals correctly. If the original request body has been consumed, you must provide a new body or clone the request before consuming.

import ky, {isForceRetryError} from 'ky';

const api = ky.extend({
	hooks: {
		afterResponse: [
			async ({request, response}) => {
				// Retry based on response body content
				if (response.status === 200) {
					const data = await response.json();

					// Simple retry with default delay
					if (data.error?.code === 'TEMPORARY_ERROR') {
						return ky.retry();
					}

					// Retry with custom delay from API response
					if (data.error?.code === 'RATE_LIMIT') {
						return ky.retry({
							delay: data.error.retryAfter * 1000,
							code: 'RATE_LIMIT'
						});
					}

					// Retry with a modified request (e.g., fallback endpoint)
					if (data.error?.code === 'FALLBACK_TO_BACKUP') {
						return ky.retry({
							request: new Request('https://backup-api.com/endpoint', {
								method: request.method,
								headers: request.headers,
							}),
							code: 'BACKUP_ENDPOINT'
						});
					}

					// Retry with refreshed authentication
					if (data.error?.code === 'TOKEN_REFRESH' && data.newToken) {
						return ky.retry({
							request: new Request(request, {
								headers: {
									...Object.fromEntries(request.headers),
									'Authorization': `Bearer ${data.newToken}`
								}
							}),
							code: 'TOKEN_REFRESHED'
						});
					}

					// Retry with cause to preserve error chain
					try {
						validateResponse(data);
					} catch (error) {
						return ky.retry({
							code: 'VALIDATION_FAILED',
							cause: error
						});
					}
				}
			}
		],
		beforeRetry: [
			({error, retryCount}) => {
				// Observable in beforeRetry hooks
				if (isForceRetryError(error)) {
					console.log(`Forced retry #${retryCount}: ${error.message}`);
					// Example output: "Forced retry #1: Forced retry: RATE_LIMIT"
				}
			}
		]
	}
});

const response = await api.get('https://example.com/api');

Base class for all Ky-specific errors. HTTPError, NetworkError, TimeoutError, ResponseSizeError, and ForceRetryError extend this class.

You can use instanceof KyError to check if an error originated from Ky, or use the isKyError() type guard for cross-realm compatibility and TypeScript type narrowing.

import ky, {isKyError} from 'ky';

try {
	await ky('https://example.com').json();
} catch (error) {
	if (isKyError(error)) {
		console.log('Ky error:', error.message);
	}
}

Exposed for instanceof checks. The error has a response property with the Response object, request property with the Request object, and options property with normalized options (either passed to ky when creating an instance with ky.create() or directly when performing the request).

It also has a data property with the pre-parsed response body. For JSON responses (based on Content-Type), the body is parsed using the parseJson option if set, or JSON.parse by default. For other content types, it is set as plain text. If the body is empty, unreadable, too large, parsing fails, or the error-data read/parse timeout is reached, data will be undefined. To avoid hanging or excessive buffering, error.data body reads and async JSON parsing are bounded by the request timeout (or 10 seconds when timeout is disabled), any remaining totalTimeout budget, and a 10 MiB response body size limit. If maxResponseSize is exceeded while populating error.data, Ky throws ResponseSizeError instead of HTTPError. If totalTimeout expires while populating error.data, Ky throws TimeoutError instead of HTTPError. The data property is populated before beforeError hooks run, so hooks can access it.

Be aware that some types of errors, such as network errors, inherently mean that a response was not received. In that case, the error will be an instance of NetworkError instead of HTTPError and will not contain a response property.

import ky, {isHTTPError} from 'ky';

try {
	await ky('https://example.com').json();
} catch (error) {
	if (isHTTPError(error)) {
		console.log(error.data);
	}
}

You can also use the beforeError hook:

import ky, {isHTTPError} from 'ky';

await ky('https://example.com', {
	hooks: {
		beforeError: [
			({error}) => {
				if (isHTTPError(error) && error.data !== undefined) {
					error.message = `${error.message}: ${JSON.stringify(error.data)}`;
				}

				return error;
			}
		]
	}
});

⌨️ TypeScript: Accepts an optional type parameter, which defaults to unknown, and is passed through to the type of error.data.

The error thrown when Standard Schema validation fails in .json(schema). It has an issues property with the validation issues from the schema.

This error intentionally does not extend KyError because it does not represent a failure in Ky's HTTP lifecycle. The request succeeded; the user's schema rejected the data. As such, it is not matched by isKyError().

import ky, {SchemaValidationError} from 'ky';
import {z} from 'zod';

const userSchema = z.object({name: z.string()});

try {
	const user = await ky('/api/user').json(userSchema);
	console.log(user.name);
} catch (error) {
	if (error instanceof SchemaValidationError) {
		console.error(error.issues);
	}
}

The error thrown when the request times out. It has a request property with the Request object.

import ky, {isTimeoutError} from 'ky';

try {
	await ky('https://example.com').json();
} catch (error) {
	if (isTimeoutError(error)) {
		console.log('Request timed out');
	}
}

Error thrown when the response body exceeds maxResponseSize. It has a request property with the Request object and a maxResponseSize property with the configured limit in bytes.

import ky, {isResponseSizeError} from 'ky';

try {
	await ky('https://example.com/data', {maxResponseSize: 1024}).json();
} catch (error) {
	if (isResponseSizeError(error)) {
		console.log(`Response exceeded ${error.maxResponseSize} bytes`);
	}
}

The error thrown when a network error occurs during the request (e.g., DNS failure, connection refused, offline). It has a request property with the Request object. The original error is available via the standard cause property.

Network errors are automatically retried (for retriable methods). A connection that drops while a Ky shortcut method like .json() is reading the response body is also wrapped in NetworkError, but it is not retried because the response has already been received.

import ky, {isNetworkError} from 'ky';

try {
	await ky('https://example.com').json();
} catch (error) {
	if (isNetworkError(error)) {
		console.log('Network error:', error.message);
		console.log('Original error:', error.cause);
	}
}

Wraps a value so that ky.extend() will replace the parent value instead of merging with it. Works with hooks, headers, search parameters, context, and any other deep-merged option.

Inside the json option, which holds your own data, use it only where it replaces an inherited value. Ky does not look for it anywhere else there. The context option is merged shallowly, so only a wrapper around the whole context is supported.

import ky, {replaceOption} from 'ky';

const api = ky.create({
	headers: {authorization: 'Bearer token', 'x-custom': 'value'},
});

// Replace all headers instead of merging
const publicApi = api.extend({
	headers: replaceOption({accept: 'application/json'}),
});
// Headers are now just {accept: 'application/json'}

Sending form data in Ky is identical to fetch. Just pass a FormData instance to the body option. The Content-Type header will be automatically set to multipart/form-data, unless you set a Content-Type in the headers option, which always takes precedence.

import ky from 'ky';

// `multipart/form-data`
const formData = new FormData();
formData.append('food', 'fries');
formData.append('drink', 'icetea');

const response = await ky.post(url, {body: formData});

If you want to send the data in application/x-www-form-urlencoded format, you will need to encode the data with URLSearchParams. Like FormData, the Content-Type is set for you unless you set one in the headers option.

import ky from 'ky';

// `application/x-www-form-urlencoded`
const searchParams = new URLSearchParams();
searchParams.set('food', 'fries');
searchParams.set('drink', 'icetea');

const response = await ky.post(url, {body: searchParams});

If you need to modify FormData in a beforeRequest hook (for example, to transform field names), delete the Content-Type header before creating a new Request:

import ky from 'ky';

const response = await ky.post(url, {
	body: formData,
	hooks: {
		beforeRequest: [
			({request}) => {
				const newFormData = new FormData();

				// Modify FormData as needed
				for (const [key, value] of formData) {
					newFormData.set(key.toLowerCase(), value);
				}

				// Delete `Content-Type` to let Request regenerate it with correct boundary
				request.headers.delete('content-type');

				return new Request(request, {body: newFormData});
			}
		]
	}
});

Ky automatically sets an appropriate Content-Type header for each request based on the data in the request body. However, some APIs require custom, non-standard content types, such as application/x-amz-json-1.1. Using the headers option, you can manually override the content type.

import ky from 'ky';

const json = await ky.post('https://example.com', {
	headers: {
		'content-type': 'application/x-amz-json-1.1'
	},
	json: {
		foo: true
	},
}).json();

console.log(json);
//=> {data: '🦄'}

Fetch (and hence Ky) has built-in support for request cancellation through the AbortController API. Read more.

Example:

import ky from 'ky';

const controller = new AbortController();
const {signal} = controller;

setTimeout(() => {
	controller.abort();
}, 5000);

try {
	console.log(await ky(url, {signal}).text());
} catch (error) {
	if (error.name === 'AbortError') {
		console.log('Fetch aborted');
	} else {
		console.error('Fetch error:', error);
	}
}

Node.js 24.5+ supports automatic proxy configuration via environment variables. Set NODE_USE_ENV_PROXY=1 or use the --use-env-proxy CLI flag.

NODE_USE_ENV_PROXY=1 HTTP_PROXY=http://proxy.example.com:8080 node app.js

Or:

node --use-env-proxy app.js

Supported environment variables:

• HTTP_PROXY / http_proxy: Proxy URL for HTTP requests
• HTTPS_PROXY / https_proxy: Proxy URL for HTTPS requests
• NO_PROXY / no_proxy: Comma-separated list of hosts to bypass the proxy

For more control, use ProxyAgent or EnvHttpProxyAgent with the dispatcher option.

import ky from 'ky';
import {ProxyAgent} from 'undici';

const proxyAgent = new ProxyAgent('http://proxy.example.com:8080');

const response = await ky('https://example.com', {
	dispatcher: proxyAgent
}).json();

Using EnvHttpProxyAgent to automatically read proxy settings from environment variables:

import ky from 'ky';
import {EnvHttpProxyAgent} from 'undici';

const proxyAgent = new EnvHttpProxyAgent();

const api = ky.extend({
	dispatcher: proxyAgent
});

const response = await api('https://example.com').json();

Undici supports HTTP/2, but it's not enabled by default. Create a custom dispatcher with the allowH2 option:

import ky from 'ky';
import {Agent, Pool} from 'undici';

const agent = new Agent({
	factory(origin, options) {
		return new Pool(origin, {
			...options,
			allowH2: true
		});
	}
});

const response = await ky('https://example.com', {
	dispatcher: agent
}).json();

Combine proxy and HTTP/2:

import ky from 'ky';
import {ProxyAgent} from 'undici';

const proxyAgent = new ProxyAgent({
	uri: 'http://proxy.example.com:8080',
	allowH2: true
});

const response = await ky('https://example.com', {
	dispatcher: proxyAgent
}).json();

You can pass a ReadableStream as the request body. Ky sets the duplex: 'half' option required by the Fetch spec for you in environments with request stream support.

import ky from 'ky';

const stream = new ReadableStream({
	start(controller) {
		controller.enqueue(new TextEncoder().encode('hello'));
		controller.close();
	},
});

const response = await ky.post('https://example.com/upload', {
	body: stream,
});

Use parse-sse:

import ky from 'ky';
import {parseServerSentEvents} from 'parse-sse';

const response = await ky('https://api.example.com/events');

for await (const event of parseServerSentEvents(response)) {
	console.log(event.data);
}

Use fetch-extras with Ky for paginating API responses:

import ky from 'ky';
import {paginate} from 'fetch-extras';

const url = 'https://api.github.com/repos/sindresorhus/ky/commits';

for await (const commit of paginate(url, {fetchFunction: ky})) {
	console.log(commit.sha);
}

Ky's TypeScript types are intentionally defined as type aliases rather than interfaces to prevent global module augmentation, which can lead to type conflicts and unexpected behavior across your codebase. If you need to add custom properties to Ky's types like KyResponse or HTTPError, create local wrapper types instead:

import ky, {HTTPError, isHTTPError} from 'ky';

interface CustomError extends HTTPError {
	customProperty: unknown;
}

// Use with type assertion
let data: unknown;

const api = ky.extend({
	hooks: {
		beforeError: [
			async ({error}) => {
				if (isHTTPError(error)) {
					(error as CustomError).customProperty = 'value';
					data = (error as CustomError).customProperty;
				}

				return error;
			}
		]
	}
});

This approach keeps your types scoped to where they're needed without polluting the global namespace.

Node.js supports fetch natively, so you can just use this package directly.

Node.js supports fetch natively, so you can use Ky directly. The main consideration is that server-side requests require absolute URLs, while client-side requests can use relative URLs. Handle this with a conditional baseUrl:

const api = ky.create({
	baseUrl: globalThis.window === undefined
		? (process.env.BASE_URL ?? 'http://localhost:3000')
		: undefined,
});

baseUrl follows standard URL resolution rules — the same behavior as new URL(input, baseUrl). A leading slash on the input means origin-root, overriding any path in the base URL.

prefix does plain string joining before URL resolution. The leading slash on the input is stripped, so it always appends to the prefix regardless.

Use baseUrl in almost all cases. Use prefix only when you want origin-relative inputs like /users to be treated as page-relative.

// On https://example.com

// baseUrl: standard URL resolution
ky('users',  {baseUrl: '/api/'});
//=> https://example.com/api/users

ky('/users', {baseUrl: '/api/'});
//=> https://example.com/users  ← leading slash wins


// prefix: always appends
ky('users',  {prefix: '/api'});
//=> https://example.com/api/users

ky('/users', {prefix: '/api'});
//=> https://example.com/api/users  ← leading slash ignored

Use a test runner that can run in the browser, like Vitest or Playwright.

Use the beforeRequest hook to attach headers before each request:

const api = ky.create({
	hooks: {
		beforeRequest: [
			({request}) => {
				request.headers.set('Authorization', `Bearer ${getToken()}`);
			}
		]
	}
});

Configure retry to retry on 401, then use the beforeRetry hook to refresh the token:

const api = ky.create({
	retry: {statusCodes: [401]},
	hooks: {
		beforeRetry: [
			async ({request}) => {
				const token = await refreshToken();
				request.headers.set('Authorization', `Bearer ${token}`);
			}
		]
	}
});

Use MSW to intercept requests at the network level without modifying your code. Alternatively, pass a custom fetch to your Ky instance:

const api = ky.create({
	fetch: async () => new Response(JSON.stringify({name: 'Test'}), {
		headers: {
			'Content-Type': 'application/json'
		},
	}),
});

Use ky.retry() in an afterResponse hook. This lets you force a retry even when the status code is 2xx:

import ky from 'ky';

const response = await ky('https://example.com', {
	hooks: {
		afterResponse: [
			async ({response}) => {
				const data = await response.json();

				if (data.status === 'pending') {
					return ky.retry();
				}
			}
		]
	}
});

The retry respects retry.limit and is observable in beforeRetry hooks.

Throw from a beforeRetry hook to stop retrying and propagate the error, or return ky.stop to stop silently (the request resolves with undefined):

import ky, {isHTTPError} from 'ky';

const response = await ky('https://example.com', {
	hooks: {
		beforeRetry: [
			({error}) => {
				if (isHTTPError(error) && error.response.status === 400) {
					throw error; // Stop retrying, propagate the error
				}
			}
		]
	}
});

For simple status-based filtering, pass a function to throwHttpErrors:

import ky from 'ky';

// Only throw on 5xx errors; 4xx responses are returned normally
const api = ky.create({
	throwHttpErrors: status => status >= 500,
});

const response = await api('https://example.com/resource');

if (response.status === 404) {
	// Handle "not found" as normal app flow
}

If you need to access the response body in the non-throwing path, use throwHttpErrors: false and throw manually in an afterResponse hook:

import ky, {HTTPError} from 'ky';

const api = ky.create({
	throwHttpErrors: false,
	hooks: {
		afterResponse: [
			({request, options, response}) => {
				if (response.status >= 500) {
					throw new HTTPError(response, request, options);
				}
			}
		]
	}
});

Make sure your code is running as a JavaScript module (ESM), for example by using a <script type="module"> tag in your HTML document. Then Ky can be imported directly by that module without a bundler or other tools.

<script type="module">
import ky from 'https://unpkg.com/ky/distribution/index.js';

const json = await ky('https://jsonplaceholder.typicode.com/todos/1').json();

console.log(json.title);
//=> 'delectus aut autem'
</script>

Got is maintained by the same people as Ky, so you probably want Ky instead. It's smaller, works in the browser too, and is more stable since it's built on Fetch.

Axios predates the Fetch API and has a lot of legacy baggage. Ky is built on Fetch, which means it's smaller, more standards-compliant, and works everywhere Fetch does (browsers, Node.js, Bun, Deno). Ky also has a more modern API with better TypeScript support.

It's just a random short npm package name I managed to get. It does, however, have a meaning in Japanese:

A form of text-able slang, KY is an abbreviation for 空気読めない (kuuki yomenai), which literally translates into “cannot read the air.” It's a phrase applied to someone who misses the implied meaning.

The latest version of Chrome, Firefox, and Safari.

Node.js 22 and later.

• fetch-extras - Useful utilities for working with Fetch
• ky-hooks-change-case - Ky hooks to modify cases on requests and responses of objects

• Sindre Sorhus
• Seth Holladay
• Szymon Marczak