Configuration Reference

The following reference covers all supported configuration options in Astro. To learn more about configuring Astro, read our guide on Configuring Astro.

import { defineConfig } from 'astro/config'

export default defineConfig({
  // your configuration options here...
})

Top-Level Options

site

Type: string

Your final, deployed URL. Astro uses this full URL to generate your sitemap and canonical URLs in your final build. It is strongly recommended that you set this configuration to get the most out of Astro.

{
  site: 'https://www.my-site.dev'
}

base

Type: string

The base path to deploy to. Astro will use this path as the root for your pages and assets both in development and in production build.

In the example below, astro dev will start your server at /docs.

{
  base: '/docs'
}

When using this option, all of your static asset imports and URLs should add the base as a prefix. You can access this value via import.meta.env.BASE_URL.

The value of import.meta.env.BASE_URL will be determined by your trailingSlash config, no matter what value you have set for base.

A trailing slash is always included if trailingSlash: "always" is set. If trailingSlash: "never" is set, BASE_URL will not include a trailing slash, even if base includes one.

Additionally, Astro will internally manipulate the configured value of config.base before making it available to integrations. The value of config.base as read by integrations will also be determined by your trailingSlash configuration in the same way.

In the example below, the values of import.meta.env.BASE_URL and config.base when processed will both be /docs:

{
   base: '/docs/',
   trailingSlash: "never"
}

In the example below, the values of import.meta.env.BASE_URL and config.base when processed will both be /docs/:

{
   base: '/docs',
   trailingSlash: "always"
}

trailingSlash

Type: 'always' | 'never' | 'ignore'
Default: 'ignore'

Set the route matching behavior of the dev server. Choose from the following options:

'always' - Only match URLs that include a trailing slash (ex: “/foo/“)
'never' - Never match URLs that include a trailing slash (ex: “/foo”)
'ignore' - Match URLs regardless of whether a trailing ”/” exists

Use this configuration option if your production host has strict handling of how trailing slashes work or do not work.

You can also set this if you prefer to be more strict yourself, so that URLs with or without trailing slashes won’t work during development.

{
  // Example: Require a trailing slash during development
  trailingSlash: 'always'
}

See Also:

build.format

redirects

Type: Record.<string, RedirectConfig>
Default: {}

Added in: astro@2.9.0

Specify a mapping of redirects where the key is the route to match and the value is the path to redirect to.

You can redirect both static and dynamic routes, but only to the same kind of route. For example you cannot have a '/article': '/blog/[...slug]' redirect.

{
  redirects: {
    '/old': '/new',
    '/blog/[...slug]': '/articles/[...slug]',
  }
}

For statically-generated sites with no adapter installed, this will produce a client redirect using a <meta http-equiv="refresh"> tag and does not support status codes.

When using SSR or with a static adapter in output: static mode, status codes are supported. Astro will serve redirected GET requests with a status of 301 and use a status of 308 for any other request method.

You can customize the redirection status code using an object in the redirect config:

{
  redirects: {
    '/other': {
      status: 302,
      destination: '/place',
    },
  }
}

output

Type: 'static' | 'server' | 'hybrid'
Default: 'static'

Specifies the output target for builds.

'static' - Building a static site to be deployed to any static host.
'server' - Building an app to be deployed to a host supporting SSR (server-side rendering).
'hybrid' - Building a static site with a few server-side rendered pages.

import { defineConfig } from 'astro/config';

export default defineConfig({
  output: 'static'
})

See Also:

adapter

adapter

Type: AstroIntegration

Deploy to your favorite server, serverless, or edge host with build adapters. Import one of our first-party adapters for Netlify, Vercel, and more to engage Astro SSR.

See our Server-side Rendering guide for more on SSR, and our deployment guides for a complete list of hosts.

import netlify from '@astrojs/netlify';
{
  // Example: Build for Netlify serverless deployment
  adapter: netlify(),
}

See Also:

output

integrations

Type: AstroIntegration[]

Extend Astro with custom integrations. Integrations are your one-stop-shop for adding framework support (like Solid.js), new features (like sitemaps), and new libraries (like Partytown).

Read our Integrations Guide for help getting started with Astro Integrations.

import react from '@astrojs/react';
import tailwind from '@astrojs/tailwind';
{
  // Example: Add React + Tailwind support to Astro
  integrations: [react(), tailwind()]
}

root

Type: string
CLI: --root
Default: "." (current working directory)

You should only provide this option if you run the astro CLI commands in a directory other than the project root directory. Usually, this option is provided via the CLI instead of the Astro config file, since Astro needs to know your project root before it can locate your config file.

If you provide a relative path (ex: --root: './my-project') Astro will resolve it against your current working directory.

Examples

{
  root: './my-project-directory'
}

$ astro build --root ./my-project-directory

srcDir

Type: string
Default: "./src"

Set the directory that Astro will read your site from.

The value can be either an absolute file system path or a path relative to the project root.

{
  srcDir: './www'
}

publicDir

Type: string
Default: "./public"

Set the directory for your static assets. Files in this directory are served at / during dev and copied to your build directory during build. These files are always served or copied as-is, without transform or bundling.

The value can be either an absolute file system path or a path relative to the project root.

{
  publicDir: './my-custom-publicDir-directory'
}

outDir

Type: string
Default: "./dist"

Set the directory that astro build writes your final build to.

The value can be either an absolute file system path or a path relative to the project root.

{
  outDir: './my-custom-build-directory'
}

See Also:

build.server

cacheDir

Type: string
Default: "./node_modules/.astro"

Set the directory for caching build artifacts. Files in this directory will be used in subsequent builds to speed up the build time.

The value can be either an absolute file system path or a path relative to the project root.

{
  cacheDir: './my-custom-cache-directory'
}

compressHTML

Type: boolean
Default: true

This is an option to minify your HTML output and reduce the size of your HTML files.

By default, Astro removes whitespace from your HTML, including line breaks, from .astro components in a lossless manner. Some whitespace may be kept as needed to preserve the visual rendering of your HTML. This occurs both in development mode and in the final build.

To disable HTML compression, set compressHTML to false.

{
  compressHTML: false
}

scopedStyleStrategy

Type: 'where' | 'class' | 'attribute'
Default: 'attribute'

Added in: astro@2.4

Specify the strategy used for scoping styles within Astro components. Choose from:

'where' - Use :where selectors, causing no specificity increase.
'class' - Use class-based selectors, causing a +1 specificity increase.
'attribute' - Use data- attributes, causing a +1 specificity increase.

Using 'class' is helpful when you want to ensure that element selectors within an Astro component override global style defaults (e.g. from a global stylesheet). Using 'where' gives you more control over specificity, but requires that you use higher-specificity selectors, layers, and other tools to control which selectors are applied. Using 'attribute' is useful when you are manipulating the class attribute of elements and need to avoid conflicts between your own styling logic and Astro’s application of styles.

security

Type: boolean
Default: {}

Added in: astro@4.9.0

Enables security measures for an Astro website.

These features only exist for pages rendered on demand (SSR) using server mode or pages that opt out of prerendering in hybrid mode.

export default defineConfig({
  output: "server",
  security: {
    checkOrigin: true
  }
})

security.checkOrigin

Type: boolean
Default: ‘false’

Added in: astro@4.9.0

When enabled, performs a check that the “origin” header, automatically passed by all modern browsers, matches the URL sent by each Request. This is used to provide Cross-Site Request Forgery (CSRF) protection.

The “origin” check is executed only for pages rendered on demand, and only for the requests POST, PATCH, DELETE and PUT with one of the following content-type headers: 'application/x-www-form-urlencoded', 'multipart/form-data', 'text/plain'.

If the “origin” header doesn’t match the pathname of the request, Astro will return a 403 status code and will not render the page.

vite

Type: ViteUserConfig

Pass additional configuration options to Vite. Useful when Astro doesn’t support some advanced configuration that you may need.

View the full vite configuration object documentation on vitejs.dev.

Examples

{
  vite: {
    ssr: {
      // Example: Force a broken package to skip SSR processing, if needed
      external: ['broken-npm-package'],
    }
  }
}

{
  vite: {
    // Example: Add custom vite plugins directly to your Astro project
    plugins: [myPlugin()],
  }
}

Build Options

build.format

Type: ('file' | 'directory' | 'preserve')
Default: 'directory'

Control the output file format of each page. This value may be set by an adapter for you.

'file': Astro will generate an HTML file named for each page route. (e.g. src/pages/about.astro and src/pages/about/index.astro both build the file /about.html)
'directory': Astro will generate a directory with a nested index.html file for each page. (e.g. src/pages/about.astro and src/pages/about/index.astro both build the file /about/index.html)
'preserve': Astro will generate HTML files exactly as they appear in your source folder. (e.g. src/pages/about.astro builds /about.html and src/pages/about/index.astro builds the file /about/index.html)

{
  build: {
    // Example: Generate `page.html` instead of `page/index.html` during build.
    format: 'file'
  }
}

Effect on Astro.url

Setting build.format controls what Astro.url is set to during the build. When it is:

directory - The Astro.url.pathname will include a trailing slash to mimic folder behavior; ie /foo/.
file - The Astro.url.pathname will include .html; ie /foo.html.

This means that when you create relative URLs using new URL('./relative', Astro.url), you will get consistent behavior between dev and build.

To prevent inconsistencies with trailing slash behaviour in dev, you can restrict the trailingSlash option to 'always' or 'never' depending on your build format:

directory - Set trailingSlash: 'always'
file - Set trailingSlash: 'never'

build.client

Type: string
Default: './dist/client'

Controls the output directory of your client-side CSS and JavaScript when output: 'server' or output: 'hybrid' only. outDir controls where the code is built to.

This value is relative to the outDir.

{
  output: 'server', // or 'hybrid'
  build: {
    client: './client'
  }
}

build.server

Type: string
Default: './dist/server'

Controls the output directory of server JavaScript when building to SSR.

This value is relative to the outDir.

{
  build: {
    server: './server'
  }
}

build.assets

Type: string
Default: '_astro'

Added in: astro@2.0.0

Specifies the directory in the build output where Astro-generated assets (bundled JS and CSS for example) should live.

{
  build: {
    assets: '_custom'
  }
}

See Also:

outDir

build.assetsPrefix

Type: string | Record.<string, string>
Default: undefined

Added in: astro@2.2.0

Specifies the prefix for Astro-generated asset links. This can be used if assets are served from a different domain than the current site.

This requires uploading the assets in your local ./dist/_astro folder to a corresponding /_astro/ folder on the remote domain. To rename the _astro path, specify a new directory in build.assets.

To fetch all assets uploaded to the same domain (e.g. https://cdn.example.com/_astro/...), set assetsPrefix to the root domain as a string (regardless of your base configuration):

{
  build: {
    assetsPrefix: 'https://cdn.example.com'
  }
}

Added in: astro@4.5.0

You can also pass an object to assetsPrefix to specify a different domain for each file type. In this case, a fallback property is required and will be used by default for any other files.

{
  build: {
    assetsPrefix: {
      'js': 'https://js.cdn.example.com',
      'mjs': 'https://js.cdn.example.com',
      'css': 'https://css.cdn.example.com',
      'fallback': 'https://cdn.example.com'
    }
  }
}

build.serverEntry

Type: string
Default: 'entry.mjs'

Specifies the file name of the server entrypoint when building to SSR. This entrypoint is usually dependent on which host you are deploying to and will be set by your adapter for you.

Note that it is recommended that this file ends with .mjs so that the runtime detects that the file is a JavaScript module.

{
  build: {
    serverEntry: 'main.mjs'
  }
}

build.redirects

Type: boolean
Default: true

Added in: astro@2.6.0

Specifies whether redirects will be output to HTML during the build. This option only applies to output: 'static' mode; in SSR redirects are treated the same as all responses.

This option is mostly meant to be used by adapters that have special configuration files for redirects and do not need/want HTML based redirects.

{
  build: {
    redirects: false
  }
}

build.inlineStylesheets

Type: 'always' | 'auto' | 'never'
Default: auto

Added in: astro@2.6.0

Control whether project styles are sent to the browser in a separate css file or inlined into <style> tags. Choose from the following options:

'always' - project styles are inlined into <style> tags
'auto' - only stylesheets smaller than ViteConfig.build.assetsInlineLimit (default: 4kb) are inlined. Otherwise, project styles are sent in external stylesheets.
'never' - project styles are sent in external stylesheets

{
  build: {
    inlineStylesheets: `never`,
  },
}

Server Options

Customize the Astro dev server, used by both astro dev and astro preview.

{
  server: { port: 1234, host: true}
}

To set different configuration based on the command run (“dev”, “preview”) a function can also be passed to this configuration option.

{
  // Example: Use the function syntax to customize based on command
  server: ({ command }) => ({ port: command === 'dev' ? 4321 : 4000 })
}

server.host

Type: string | boolean
Default: false

Added in: astro@0.24.0

Set which network IP addresses the server should listen on (i.e. non-localhost IPs).

false - do not expose on a network IP address
true - listen on all addresses, including LAN and public addresses
[custom-address] - expose on a network IP address at [custom-address] (ex: 192.168.0.1)

server.port

Type: number
Default: 4321

Set which port the server should listen on.

If the given port is already in use, Astro will automatically try the next available port.

{
  server: { port: 8080 }
}

server.open

Type: string | boolean
Default: false

Added in: astro@4.1.0

Controls whether the dev server should open in your browser window on startup.

Pass a full URL string (e.g. ”http://example.com”) or a pathname (e.g. “/about”) to specify the URL to open.

{
  server: { open: "/about" }
}

server.headers

Type: OutgoingHttpHeaders
Default: {}

Added in: astro@1.7.0

Set custom HTTP response headers to be sent in astro dev and astro preview.

devToolbar.enabled

Type: boolean
Default: true

Whether to enable the Astro Dev Toolbar. This toolbar allows you to inspect your page islands, see helpful audits on performance and accessibility, and more.

This option is scoped to the entire project, to only disable the toolbar for yourself, run npm run astro preferences disable devToolbar. To disable the toolbar for all your Astro projects, run npm run astro preferences disable devToolbar --global.

Prefetch Options

Type: boolean | object

Enable prefetching for links on your site to provide faster page transitions. (Enabled by default on pages using the <ViewTransitions /> router. Set prefetch: false to opt out of this behaviour.)

This configuration automatically adds a prefetch script to every page in the project giving you access to the data-astro-prefetch attribute. Add this attribute to any <a /> link on your page to enable prefetching for that page.

<a href="/about" data-astro-prefetch>About</a>

Further customize the default prefetching behavior using the prefetch.defaultStrategy and prefetch.prefetchAll options.

See the Prefetch guide for more information.

prefetch.prefetchAll

Type: boolean

Enable prefetching for all links, including those without the data-astro-prefetch attribute. This value defaults to true when using the <ViewTransitions /> router. Otherwise, the default value is false.

prefetch: {
  prefetchAll: true
}

When set to true, you can disable prefetching individually by setting data-astro-prefetch="false" on any individual links.

<a href="/about" data-astro-prefetch="false">About</a>

prefetch.defaultStrategy

Type: 'tap' | 'hover' | 'viewport' | 'load'
Default: 'hover'

The default prefetch strategy to use when the data-astro-prefetch attribute is set on a link with no value.

'tap': Prefetch just before you click on the link.
'hover': Prefetch when you hover over or focus on the link. (default)
'viewport': Prefetch as the links enter the viewport.
'load': Prefetch all links on the page after the page is loaded.

You can override this default value and select a different strategy for any individual link by setting a value on the attribute.

<a href="/about" data-astro-prefetch="viewport">About</a>

Image Options

image.endpoint

Type: string
Default: undefined

Added in: astro@3.1.0

Set the endpoint to use for image optimization in dev and SSR. Set to undefined to use the default endpoint.

The endpoint will always be injected at /_image.

{
  image: {
    // Example: Use a custom image endpoint
    endpoint: './src/image-endpoint.ts',
  },
}

image.service

Type: Object
Default: {entrypoint: 'astro/assets/services/sharp', config?: {}}

Added in: astro@2.1.0

Set which image service is used for Astro’s assets support.

The value should be an object with an entrypoint for the image service to use and optionally, a config object to pass to the service.

The service entrypoint can be either one of the included services, or a third-party package.

{
  image: {
    // Example: Enable the Sharp-based image service with a custom config
    service: {
       entrypoint: 'astro/assets/services/sharp',
       config: {
         limitInputPixels: false,
      },
     },
  },
}

image.service.config.limitInputPixels

Type: number | boolean
Default: true

Added in: astro@4.1.0

Whether or not to limit the size of images that the Sharp image service will process.

Set false to bypass the default image size limit for the Sharp image service and process large images.

image.domains

Type: Array.<string>
Default: {domains: []}

Added in: astro@2.10.10

Defines a list of permitted image source domains for remote image optimization. No other remote images will be optimized by Astro.

This option requires an array of individual domain names as strings. Wildcards are not permitted. Instead, use image.remotePatterns to define a list of allowed source URL patterns.

{
  image: {
    // Example: Allow remote image optimization from a single domain
    domains: ['astro.build'],
  },
}

image.remotePatterns

Type: Array.<RemotePattern>
Default: {remotePatterns: []}

Added in: astro@2.10.10

Defines a list of permitted image source URL patterns for remote image optimization.

remotePatterns can be configured with four properties:

protocol
hostname
port
pathname

{
  image: {
    // Example: allow processing all images from your aws s3 bucket
    remotePatterns: [{
      protocol: 'https',
      hostname: '**.amazonaws.com',
    }],
  },
}

You can use wildcards to define the permitted hostname and pathname values as described below. Otherwise, only the exact values provided will be configured: hostname:

Start with ’**.’ to allow all subdomains (‘endsWith’).
Start with ’*.’ to allow only one level of subdomain.

pathname:

End with ’/**’ to allow all sub-routes (‘startsWith’).
End with ’/*’ to allow only one level of sub-route.

Markdown Options

markdown.shikiConfig

Type: Partial<ShikiConfig>

Shiki configuration options. See the Markdown configuration docs for usage.

markdown.syntaxHighlight

Type: 'shiki' | 'prism' | false
Default: shiki

Which syntax highlighter to use, if any.

shiki - use the Shiki highlighter
prism - use the Prism highlighter
false - do not apply syntax highlighting.

{
  markdown: {
    // Example: Switch to use prism for syntax highlighting in Markdown
    syntaxHighlight: 'prism',
  }
}

markdown.remarkPlugins

Type: RemarkPlugins

Pass remark plugins to customize how your Markdown is built. You can import and apply the plugin function (recommended), or pass the plugin name as a string.

import remarkToc from 'remark-toc';
{
  markdown: {
    remarkPlugins: [ [remarkToc, { heading: "contents"} ] ]
  }
}

markdown.rehypePlugins

Type: RehypePlugins

Pass rehype plugins to customize how your Markdown’s output HTML is processed. You can import and apply the plugin function (recommended), or pass the plugin name as a string.

import { rehypeAccessibleEmojis } from 'rehype-accessible-emojis';
{
  markdown: {
    rehypePlugins: [rehypeAccessibleEmojis]
  }
}

markdown.gfm

Type: boolean
Default: true

Added in: astro@2.0.0

Astro uses GitHub-flavored Markdown by default. To disable this, set the gfm flag to false:

{
  markdown: {
    gfm: false,
  }
}

markdown.smartypants

Type: boolean
Default: true

Added in: astro@2.0.0

Astro uses the SmartyPants formatter by default. To disable this, set the smartypants flag to false:

{
  markdown: {
    smartypants: false,
  }
}

markdown.remarkRehype

Type: RemarkRehype

Pass options to remark-rehype.

{
  markdown: {
    // Example: Translate the footnotes text to another language, here are the default English values
    remarkRehype: { footnoteLabel: "Footnotes", footnoteBackLabel: "Back to reference 1"},
  },
};

i18n

Type: object

Added in: astro@3.5.0

Configures i18n routing and allows you to specify some customization options.

See our guide for more information on internationalization in Astro

i18n.defaultLocale

Type: string

Added in: astro@3.5.0

The default locale of your website/application. This is a required field.

No particular language format or syntax is enforced, but we suggest using lower-case and hyphens as needed (e.g. “es”, “pt-br”) for greatest compatibility.

i18n.locales

Type: Locales

Added in: astro@3.5.0

A list of all locales supported by the website, including the defaultLocale. This is a required field.

Languages can be listed either as individual codes (e.g. ['en', 'es', 'pt-br']) or mapped to a shared path of codes (e.g. { path: "english", codes: ["en", "en-US"]}). These codes will be used to determine the URL structure of your deployed site.

No particular language code format or syntax is enforced, but your project folders containing your content files must match exactly the locales items in the list. In the case of multiple codes pointing to a custom URL path prefix, store your content files in a folder with the same name as the path configured.

i18n.fallback

Type: Record.<string, string>

Added in: astro@3.5.0

The fallback strategy when navigating to pages that do not exist (e.g. a translated page has not been created).

Use this object to declare a fallback locale route for each language you support. If no fallback is specified, then unavailable pages will return a 404.

Example

The following example configures your content fallback strategy to redirect unavailable pages in /pt-br/ to their es version, and unavailable pages in /fr/ to their en version. Unavailable /es/ pages will return a 404.

export default defineConfig({
  i18n: {
    defaultLocale: "en",
    locales: ["en", "fr", "pt-br", "es"],
    fallback: {
      pt: "es",
      fr: "en"
    }
  }
})

i18n.routing

Type: Routing

Added in: astro@3.7.0

Controls the routing strategy to determine your site URLs. Set this based on your folder/URL path configuration for your default language.

i18n.routing.prefixDefaultLocale

Type: boolean
Default: false

Added in: astro@3.7.0

When false, only non-default languages will display a language prefix. The defaultLocale will not show a language prefix and content files do not exist in a localized folder. URLs will be of the form example.com/[locale]/content/ for all non-default languages, but example.com/content/ for the default locale.

When true, all URLs will display a language prefix. URLs will be of the form example.com/[locale]/content/ for every route, including the default language. Localized folders are used for every language, including the default.

export default defineConfig({
  i18n: {
    defaultLocale: "en",
    locales: ["en", "fr", "pt-br", "es"],
    routing: {
      prefixDefaultLocale: true,
    }
  }
})

i18n.routing.redirectToDefaultLocale

Type: boolean
Default: true

Added in: astro@4.2.0

Configures whether or not the home URL (/) generated by src/pages/index.astro will redirect to /[defaultLocale] when prefixDefaultLocale: true is set.

Set redirectToDefaultLocale: false to disable this automatic redirection at the root of your site:

export default defineConfig({
  i18n:{
    defaultLocale: "en",
    locales: ["en", "fr"],
    routing: {
      prefixDefaultLocale: true,
      redirectToDefaultLocale: false
    }
  }
})

i18n.routing.manual

Type: string

Added in: astro@4.6.0

When this option is enabled, Astro will disable its i18n middleware so that you can implement your own custom logic. No other routing options (e.g. prefixDefaultLocale) may be configured with routing: "manual".

You will be responsible for writing your own routing logic, or executing Astro’s i18n middleware manually alongside your own.

export default defineConfig({
  i18n: {
    defaultLocale: "en",
    locales: ["en", "fr", "pt-br", "es"],
    routing: {
      prefixDefaultLocale: true,
    }
  }
})

Legacy Flags

To help some users migrate between versions of Astro, we occasionally introduce legacy flags. These flags allow you to opt in to some deprecated or otherwise outdated behavior of Astro in the latest version, so that you can continue to upgrade and take advantage of new Astro releases.

Experimental Flags

Astro offers experimental flags to give users early access to new features. These flags are not guaranteed to be stable.

experimental.directRenderScript

Type: boolean
Default: false

Added in: astro@4.5.0

Enables a more reliable strategy to prevent scripts from being executed in pages where they are not used.

Scripts will directly render as declared in Astro files (including existing features like TypeScript, importing node_modules, and deduplicating scripts). You can also now conditionally render scripts in your Astro file. However, this means scripts are no longer hoisted to the <head> and multiple scripts on a page are no longer bundled together. If you enable this option, you should check that all your <script> tags behave as expected.

This option will be enabled by default in Astro 5.0.

{
  experimental: {
    directRenderScript: true,
  },
}

experimental.actions

Type: boolean
Default: false

Added in: astro@4.8.0

Actions help you write type-safe backend functions you can call from anywhere. Enable server rendering using the output property and add the actions flag to the experimental object:

{
  output: 'hybrid', // or 'server'
  experimental: {
    actions: true,
  },
}

Declare all your actions in src/actions/index.ts. This file is the global actions handler.

Define an action using the defineAction() utility from the astro:actions module. An action accepts the handler property to define your server-side request handler. If your action accepts arguments, apply the input property to validate parameters with Zod.

This example defines two actions: like and comment. The like action accepts a JSON object with a postId string, while the comment action accepts FormData with postId, author, and body strings. Each handler updates your database and return a type-safe response.

import { defineAction, z } from "astro:actions";

export const server = {
  like: defineAction({
    input: z.object({ postId: z.string() }),
    handler: async ({ postId }) => {
      // update likes in db

      return likes;
    },
  }),
  comment: defineAction({
    accept: 'form',
    input: z.object({
      postId: z.string(),
      author: z.string(),
      body: z.string(),
    }),
    handler: async ({ postId }) => {
      // insert comments in db

      return comment;
    },
  }),
};

Then, call an action from your client components using the actions object from astro:actions. You can pass a type-safe object when using JSON, or a FormData object when using accept: 'form' in your action definition.

This example calls the like and comment actions from a React component:

import { actions } from "astro:actions";
import { useState } from "react";

export function Like({ postId }: { postId: string }) {
  const [likes, setLikes] = useState(0);
  return (
    <button
      onClick={async () => {
        const newLikes = await actions.like({ postId });
        setLikes(newLikes);
      }}
    >
      {likes} likes
    </button>
  );
}

export function Comment({ postId }: { postId: string }) {
  return (
    <form
      onSubmit={async (e) => {
        e.preventDefault();
        const formData = new FormData(e.target as HTMLFormElement);
        const result = await actions.blog.comment(formData);
        // handle result
      }}
    >
      <input type="hidden" name="postId" value={postId} />
      <label htmlFor="author">Author</label>
      <input id="author" type="text" name="author" />
      <textarea rows={10} name="body"></textarea>
      <button type="submit">Post</button>
    </form>
  );
}

For a complete overview, and to give feedback on this experimental API, see the Actions RFC.

experimental.contentCollectionCache

Type: boolean
Default: false

Added in: astro@3.5.0

Enables a persistent cache for content collections when building in static mode.

{
  experimental: {
    contentCollectionCache: true,
  },
}

experimental.clientPrerender

Type: boolean
Default: false

Added in: astro@4.2.0

Enables pre-rendering your prefetched pages on the client in supported browsers.

This feature uses the experimental Speculation Rules Web API and enhances the default prefetch behavior globally to prerender links on the client. You may wish to review the possible risks when prerendering on the client before enabling this feature.

Enable client side prerendering in your astro.config.mjs along with any desired prefetch configuration options:

{
  prefetch: {
    prefetchAll: true,
    defaultStrategy: 'viewport',
  },
  experimental: {
    clientPrerender: true,
  },
}

Continue to use the data-astro-prefetch attribute on any <a /> link on your site to opt in to prefetching. Instead of appending a <link> tag to the head of the document or fetching the page with JavaScript, a <script> tag will be appended with the corresponding speculation rules.

Client side prerendering requires browser support. If the Speculation Rules API is not supported, prefetch will fallback to the supported strategy.

See the Prefetch Guide for more prefetch options and usage.

experimental.globalRoutePriority

Type: boolean
Default: false

Added in: astro@4.2.0

Prioritizes redirects and injected routes equally alongside file-based project routes, following the same route priority order rules for all routes.

This allows more control over routing in your project by not automatically prioritizing certain types of routes, and standardizes the route priority ordering for all routes.

The following example shows which route will build certain page URLs when file-based routes, injected routes, and redirects are combined as shown below:

File-based route: /blog/post/[pid]
File-based route: /[page]
Injected route: /blog/[...slug]
Redirect: /blog/tags/[tag] -> /[tag]
Redirect: /posts -> /blog

With experimental.globalRoutingPriority enabled (instead of Astro 4.0 default route priority order):

/blog/tags/astro is built by the redirect to /tags/[tag] (instead of the injected route /blog/[...slug])
/blog/post/0 is built by the file-based route /blog/post/[pid] (instead of the injected route /blog/[...slug])
/posts is built by the redirect to /blog (instead of the file-based route /[page])

In the event of route collisions, where two routes of equal route priority attempt to build the same URL, Astro will log a warning identifying the conflicting routes.

experimental.env

Type: object
Default: undefined

Added in: astro@4.10.0

Enables experimental astro:env features.

The astro:env API lets you configure a type-safe schema for your environment variables, and indicate whether they should be available on the server or the client. Import and use your defined variables from the appropriate /client or /server module:

---
import { API_URL } from "astro:env/client"
import { API_SECRET_TOKEN } from "astro:env/server"

const data = await fetch(`${API_URL}/users`, {
  method: "GET",
  headers: {
    "Content-Type": "application/json",
    "Authorization": `Bearer ${API_SECRET_TOKEN}`
  },
})
---

<script>
import { API_URL } from "astro:env/client"

fetch(`${API_URL}/ping`)
</script>

To define the data type and properties of your environment variables, declare a schema in your Astro config in experimental.env.schema. The envField helper allows you define your variable as a string, number, or boolean and pass properties in an object:

import { defineConfig, envField } from "astro/config"

export default defineConfig({
    experimental: {
        env: {
            schema: {
                API_URL: envField.string({ context: "client", access: "public", optional: true }),
                PORT: envField.number({ context: "server", access: "public", default: 4321 }),
                API_SECRET: envField.string({ context: "server", access: "secret" }),
            }
        }
    }
})

There are currently four data types supported: strings, numbers, booleans and enums.

There are three kinds of environment variables, determined by the combination of context (client or server) and access (secret or public) settings defined in your env.schema:

Public client variables: These variables end up in both your final client and server bundles, and can be accessed from both client and server through the astro:env/client module:
```
import { API_URL } from "astro:env/client"
```
Public server variables: These variables end up in your final server bundle and can be accessed on the server through the astro:env/server module:
```
import { PORT } from "astro:env/server"
```
Secret server variables: These variables are not part of your final bundle and can be accessed on the server through the astro:env/server module. The getSecret() helper function can be used to retrieve secrets not specified in the schema. Its implementation is provided by your adapter and defaults to process.env:
```
import { API_SECRET, getSecret } from "astro:env/server"

const SECRET_NOT_IN_SCHEMA = getSecret("SECRET_NOT_IN_SCHEMA") // string | undefined
```

Note: Secret client variables are not supported because there is no safe way to send this data to the client. Therefore, it is not possible to configure both context: "client" and access: "secret" in your schema.

For a complete overview, and to give feedback on this experimental API, see the Astro Env RFC.

experimental.env.schema

Type: EnvSchema
Default: undefined

Added in: astro@4.10.0

An object that uses envField to define the data type (string, number, or boolean) and properties of your environment variables: context (client or server), access (public or secret), a default value to use, and whether or not this environment variable is optional (defaults to false).

import { defineConfig, envField } from "astro/config"

export default defineConfig({
  experimental: {
    env: {
      schema: {
        API_URL: envField.string({ context: "client", access: "public", optional: true }),
        PORT: envField.number({ context: "server", access: "public", default: 4321 }),
        API_SECRET: envField.string({ context: "server", access: "secret" }),
      }
    }
  }
})

experimental.env.validateSecrets

Type: boolean
Default: false

Added in: astro@4.11.6

Whether or not to validate secrets on the server when starting the dev server or running a build.

By default, only public variables are validated on the server when starting the dev server or a build, and private variables are validated at runtime only. If enabled, private variables will also be checked on start. This is useful in some continuous integration (CI) pipelines to make sure all your secrets are correctly set before deploying.

import { defineConfig, envField } from "astro/config"

export default defineConfig({
  experimental: {
    env: {
      schema: {
        // ...
      },
      validateSecrets: true
    }
  }
})

experimental.serverIslands

Type: boolean
Default: false

Added in: astro@4.12.0

Enables experimental Server Island features. Server Islands offer the ability to defer a component to render asynchronously after the page has already rendered.

To enable, configure an on-demand server rendering output mode with an adapter, and add the serverIslands flag to the experimental object:

{
  output: 'hybrid', // or 'server'
  adapter: nodejs({ mode: 'standalone' }),
  experimental: {
    serverIslands: true,
  },
}

Use the server:defer directive on any Astro component to delay initial rendering:

---
import Avatar from '~/components/Avatar.astro';
---
<Avatar server:defer />

The outer page will be rendered, either at build time (hybrid) or at runtime (server) with the island content omitted and a <script> tag included in its place.

After the page loads in the browser, the script tag will replace itself with the the contents of the island by making a request.

Any Astro component can be given the server: defer attribute to delay its rendering. There is no special API and you can write .astro code as normal:

---
import { getUser } from '../api';

const user = await getUser(Astro.locals.userId);
---
<img class="avatar" src={user.imageUrl}>

Server island fallback content

Since your component will not render with the rest of the page, you may want to add generic content (e.g. a loading message) to temporarily show in its place. This content will be displayed when the page first renders but before the island has loaded.

Add placeholder content as a child of your Astro component with the slot="fallback" attribute. When your island content is available, the fallback content will be replaced.

The example below displays a generic avatar as fallback content, then animates into a personalized avatar using view transitions:

<Avatar server:defer>
  <svg slot="fallback" class="generic-avatar" transition:name="avatar">...</svg>
</Avatar>

For a complete overview, and to give feedback on this experimental API, see the Server Islands RFC.

experimental.contentIntellisense

Type: boolean
Default: false

Added in: astro@4.14.0 New

Enables Intellisense features (e.g. code completion, quick hints) for your content collection entries in compatible editors.

When enabled, this feature will generate and add JSON schemas to the .astro directory in your project. These files can be used by the Astro language server to provide Intellisense inside content files (.md, .mdx, .mdoc).

{
  experimental: {
    contentIntellisense: true,
  },
}

To use this feature with the Astro VS Code extension, you must also enable the astro.content-intellisense option in your VS Code settings. For editors using the Astro language server directly, pass the contentIntellisense: true initialization parameter to enable this feature.

experimental.contentLayer

Type: boolean
Default: false

Added in: astro@4.14.0 New

The Content Layer API is a new way to handle content and data in Astro. It is similar to and builds upon content collections, taking them beyond local files in src/content/ and allowing you to fetch content from anywhere, including remote APIs, by adding a loader to your collection.

Your existing content collections can be migrated to the Content Layer API with a few small changes. However, it is not necessary to update all your collections at once to add a new collection powered by the Content Layer API. You may have collections using both the existing and new APIs defined in src/content/config.ts at the same time.

The Content Layer API is designed to be more powerful and more performant, helping sites scale to thousands of pages. Data is cached between builds and updated incrementally. Markdown parsing is also 5-10 times faster, with similar scale reductions in memory, and MDX is 2-3 times faster.

To enable, add the contentLayer flag to the experimental object in your Astro config:

{
  experimental: {
    contentLayer: true,
  }
}

Fetching data with a `loader`

The Content Layer API allows you to fetch your content from outside of the src/content/ folder (whether stored locally in your project or remotely) and uses a loader property to retrieve your data.

The loader is defined in the collection’s schema and returns an array of entries. Astro provides two built-in loader functions (glob() and file()) for fetching your local content, as well as access to the API to construct your own loader and fetch remote data.

The glob() loader creates entries from directories of Markdown, MDX, Markdoc, or JSON files from anywhere on the filesystem. It accepts a pattern of entry files to match, and a base file path of where your files are located. Use this when you have one file per entry.

The file() loader creates multiple entries from a single local file. Use this when all your entries are stored in an array of objects.

import { defineCollection, z } from 'astro:content';
import { glob, file } from 'astro/loaders';

const blog = defineCollection({
  // By default the ID is a slug generated from
  // the path of the file relative to `base`
  loader: glob({ pattern: "**\/*.md", base: "./src/data/blog" }),
  schema: z.object({
    title: z.string(),
    description: z.string(),
    pubDate: z.coerce.date(),
    updatedDate: z.coerce.date().optional(),
  })
});

const dogs = defineCollection({
  // The path is relative to the project root, or an absolute path.
  loader: file("src/data/dogs.json"),
  schema: z.object({
    id: z.string(),
    breed: z.string(),
    temperament: z.array(z.string()),
  }),
});

export const collections = { blog, dogs };

Querying and rendering with the Content Layer API

The collection can be queried in the same way as content collections:

import { getCollection, getEntry } from 'astro:content';

// Get all entries from a collection.
// Requires the name of the collection as an argument.
const allBlogPosts = await getCollection('blog');

// Get a single entry from a collection.
// Requires the name of the collection and ID
const labradorData = await getEntry('dogs', 'labrador-retriever');

Entries generated from Markdown, MDX, or Markdoc can be rendered directly to a page using the render() function.

---
import { getEntry, render } from 'astro:content';

const post = await getEntry('blog', Astro.params.slug);

const { Content, headings } = await render(post);
---

<Content />

Creating a loader

With the Content Layer API, you can build loaders to load or generate content from anywhere.

For example, you can create a loader that fetches collection entries from a remote API.

const countries = defineCollection({
  loader: async () => {
    const response = await fetch("https://restcountries.com/v3.1/all");
    const data = await response.json();
    // Must return an array of entries with an id property,
    // or an object with IDs as keys and entries as values
    return data.map((country) => ({
      id: country.cca3,
      ...country,
    }));
  },
  // optionally add a schema
  // schema: z.object...
});

export const collections = { countries };

For more advanced loading logic, you can define an object loader. This allows incremental updates and conditional loading while also giving full access to the data store. See the API in the Content Layer API RFC.

Migrating an existing content collection to use the Content Layer API

You can convert an existing content collection with Markdown, MDX, Markdoc, or JSON entries to use the Content Layer API.

Move the collection folder out of src/content/ (e.g. to src/data/). All collections located in the src/content/ folder will use the existing Content Collections API.

Do not move the existing src/content/config.ts file. This file will define all collections, using either API.

Edit the collection definition. Your updated collection requires a loader, and the option to select a collection type is no longer available.

import { defineCollection, z } from 'astro:content';
import { glob } from 'astro/loaders';

const blog = defineCollection({
  // For content layer you no longer define a `type`
  type: 'content',
  loader: glob({ pattern: '**\/[^_]*.md', base: "./src/data/blog" }),
  schema: z.object({
    title: z.string(),
    description: z.string(),
    pubDate: z.coerce.date(),
    updatedDate: z.coerce.date().optional(),
  }),
});

Change references from slug to id. Content layer collections do not have a slug field. Instead, all updated collections will have an id.

---
export async function getStaticPaths() {
  const posts = await getCollection('blog');
  return posts.map((post) => ({
    params: { slug: post.slug },
    params: { slug: post.id },
    props: post,
  }));
}
---

Switch to the new render() function. Entries no longer have a render() method, as they are now serializable plain objects. Instead, import the render() function from astro:content.

---
import { getEntry } from 'astro:content';
import { getEntry, render } from 'astro:content';

const post = await getEntry('blog', params.slug);

const { Content, headings } = await post.render();
const { Content, headings } = await render(post);
---

<Content />

Learn more

For a complete overview and the full API reference, see the Content Layer API RFC and share your feedback.

Contribute

Community

Configuration Reference

Top-Level Options

site

base

trailingSlash

redirects

output

adapter

integrations

root

Examples

srcDir

publicDir

outDir

cacheDir

compressHTML

scopedStyleStrategy

security

security.checkOrigin

vite

Examples

Build Options

build.format

Effect on Astro.url

build.client

build.server

build.assets

build.assetsPrefix

build.serverEntry

build.redirects

build.inlineStylesheets

Server Options

server.host

server.port

server.open

server.headers

Dev Toolbar Options

devToolbar.enabled

Prefetch Options

prefetch.prefetchAll

prefetch.defaultStrategy

Image Options

image.endpoint

image.service

image.service.config.limitInputPixels

image.domains

image.remotePatterns

Markdown Options

markdown.shikiConfig

markdown.syntaxHighlight

markdown.remarkPlugins

markdown.rehypePlugins

markdown.gfm

markdown.smartypants

markdown.remarkRehype

i18n

i18n.defaultLocale

i18n.locales

i18n.fallback

Example

i18n.routing

i18n.routing.prefixDefaultLocale

i18n.routing.redirectToDefaultLocale

i18n.routing.manual

Legacy Flags

Experimental Flags

experimental.directRenderScript

experimental.actions

experimental.contentCollectionCache

experimental.clientPrerender

experimental.globalRoutePriority

experimental.env

experimental.env.schema

experimental.env.validateSecrets

experimental.serverIslands

Server island fallback content

experimental.contentIntellisense

experimental.contentLayer

Fetching data with a loader

Querying and rendering with the Content Layer API

Fetching data with a `loader`