Load configuration variables from multiple sources type safely using Zod.
The new major version supports both Zod 4 and Zod 3 out of the box. Check the compatibility section for more details.
- 👮‍♂️ Type safety. Zod Config uses Zod;
- 🤌 Tiny. Zod Config is a tiny library with no dependencies, tree-shaking friendly;
- ✨ Flexible. Combine multiple adapters to load the configuration from different sources. We deeply merge the configuration from different sources, following the order of the adapters provided; Create your own adapters easily; Use the callback functions to handle errors and success due to the async nature of the adapters;
- 🪴 Easy to use. Zod Config is designed to be easy to use, with a simple API;
- 🔄 Async / Sync support. Zod Config provides both asynchronous and synchronous APIs to fit different application needs;
npm install zod-config zod # npm
pnpm add zod-config zod # pnpm
yarn add zod-config zod # yarn
- Quick Start
- Compatibility
- Basic Usage
- Built In Adapters
- Advanced Features
- Contributing notes
- On the web
Zod Config provides both asynchronous and synchronous APIs for loading configuration:
loadConfig
: Asynchronous function that takes a Zod Object schema and returns a promise resolving to the validated configuration object;loadConfigSync
: Synchronous version that takes the same configuration but returns the result directly without a promise;
Note:
loadConfigSync
only supports synchronous adapters and schemas. For details on compatibility, see Synchronous loading.
Here are the available configuration options:
Here are the available configuration options for loadConfig
and loadConfigSync
:
Property | Type | Description | Required | Global | Adapter |
---|---|---|---|---|---|
schema |
AnyZodObject |
A Zod Object schema to validate the configuration. | âś… | N/A | N/A |
adapters |
Array<Adapter | SyncAdapter> | Adapter | SyncAdapter |
Adapter(s) to load the configuration from. If not provided, process.env will be used. | ❌ | N/A | N/A |
onError |
(error: InferredErrorConfig<T>) => void) |
A callback to be called when an error occurs. | ❌ | ✅ | ❌ |
onSuccess |
(data: InferredDataConfig<S>) => void |
A callback to be called when the configuration is loaded successfully. | ❌ | ✅ | ❌ |
logger |
Logger |
A custom logger to be used to log messages. By default, it uses console . |
❌ | ✅ | ❌ |
keyMatching |
'strict' | 'lenient' |
How to match keys between the schema and the data of the adapters. By default, it uses strict . |
❌ | ✅ | ✅ |
silent |
boolean |
Whether to suppress errors. By default, it is false . |
❌ | ✅ | ✅ |
transform |
(obj: { key: string; value: unknown }) => { key: string; value: unknown } | false |
Function to transform key-value pairs before processing. If the function returns false , the key-value pair will be dropped. |
❌ | ✅ | ✅ |
Note: Options marked as both "Global" and "Adapter" can be set at the global level (affecting all adapters) or at individual adapter level (affecting only that adapter). When both are provided, the adapter-level option takes precedence. For specific adapter options, check the section of the adapter you are using.
From the package we also expose the necessary types in case you want to use them in your own adapters. Some of the options are shared between the global config and the adapter config, so you can use them in your own adapters as well.
This library provides some built in adapters to load the configuration from different sources via modules. You can easily import them from zod-config/<built-in-adapter-module-name>
.
Zod Config supports both Zod 4 and Zod 3 out of the box. To start using it, just make sure you have the correct versions of zod-config
(zod-config@^1.0.0) and zod
(zod@^3.25.0 or zod@^4.0.0)!
// Using Zod 4
import { z } from "zod/v4"; // for ^4.0.0 zod versions, you can import from "zod" instead
// Using Zod 4 Mini
import { z } from "zod/v4-mini"; // for ^4.0.0 zod versions, you can import from "zod/mini" instead
// Using Zod 3
import { z } from "zod/v3";
import { loadConfig } from "zod-config";
import { envAdapter } from "zod-config/env-adapter";
const schema = z.object({
name: z.string(),
});
const config = await loadConfig({
schema,
adapters: [
envAdapter(),
],
});
By default, Zod Config will load the configuration from process.env
, no need to provide any adapter.
import { z } from 'zod';
import { loadConfig } from 'zod-config';
const schemaConfig = z.object({
port: z.string().regex(/^\d+$/),
host: z.string(),
});
const config = await loadConfig({
schema: schemaConfig
});
// config is now type safe!
console.log(config.port)
console.log(config.host)
Loads the configuration from process.env
or a custom object, allowing you to filter the keys using a regex
(this can be useful when you have multiple adapters and you want to filter the keys to avoid conflicts or just to keep only the keys you need to process). To support nested objects, you can use the nestingSeparator
property that will be used to create nested objects from flat keys based on the separator.
import { z } from 'zod';
import { loadConfig } from 'zod-config';
import { envAdapter } from 'zod-config/env-adapter';
const schemaConfig = z.object({
MY_APP_PORT: z.string().regex(/^\d+$/),
MY_APP_HOST: z.string(),
});
// using default env (process.env)
const config = await loadConfig({
schema: schemaConfig,
adapters: envAdapter(),
});
// using custom env + filter regex to match only the keys we need
const customConfig = await loadConfig({
schema: schemaConfig,
adapters: envAdapter({
regex: /^MY_APP_/,
customEnv: {
MY_APP_PORT: '3000',
MY_APP_HOST: 'localhost',
IGNORED_KEY: 'ignored',
}})
});
// using nesting separator to create nested objects
const nestedConfig = await loadConfig({
schema: z.object({
database: z.object({
host: z.string(),
port: z.string(),
}),
}),
adapters: envAdapter({
customEnv: {
'database.host': 'localhost',
'database.port': '5432',
},
nestingSeparator: '.',
}),
});
Loads the configuration from a json
file.
import { z } from 'zod';
import { loadConfig } from 'zod-config';
import { jsonAdapter } from 'zod-config/json-adapter';
import path from 'path';
const schemaConfig = z.object({
MY_APP_PORT: z.string().regex(/^\d+$/),
MY_APP_HOST: z.string(),
});
const filePath = path.join(__dirname, 'config.json');
const config = await loadConfig({
schema: schemaConfig,
adapters: jsonAdapter({ path: filePath }),
});
// using filter regex to match only the keys we need
const customConfig = await loadConfig({
schema: schemaConfig,
adapters: jsonAdapter({
path: filePath,
regex: /^MY_APP_/,
}),
});
Loads the configuration from a json5
file. In order to use this adapter, you need to install json5
(peer dependency), if you don't have it already.
import { z } from 'zod';
import { loadConfig } from 'zod-config';
import { json5Adapter } from 'zod-config/json5-adapter';
import path from 'path';
/** content of config.json5
{
// Server settings
host: 'localhost', // Single quotes + unquoted key
port: 3000, // Trailing comma
allowedIPs: [
'192.168.0.1',
'10.0.0.1', // Local access
]
}
*/
const filePath = path.join(__dirname, 'config.json5');
const schemaConfig = z.object({
host: z.string(),
port: z.number(),
allowedIPs: z.array(z.string()),
});
const config = await loadConfig({
schema: schemaConfig,
adapters: json5Adapter({ path: filePath }),
});
Loads the configuration from a yaml
file. In order to use this adapter, you need to install yaml
(peer dependency), if you don't have it already.
npm install yaml
import { z } from 'zod';
import { loadConfig } from 'zod-config';
import { yamlAdapter } from 'zod-config/yaml-adapter';
import path from 'path';
const schemaConfig = z.object({
MY_APP_PORT: z.string().regex(/^\d+$/),
MY_APP_HOST: z.string(),
});
const filePath = path.join(__dirname, 'config.yaml');
const config = await loadConfig({
schema: schemaConfig,
adapters: yamlAdapter({ path: filePath }),
});
// using filter regex to match only the keys we need
const customConfig = await loadConfig({
schema: schemaConfig,
adapters: yamlAdapter({
path: filePath,
regex: /^MY_APP_/,
}),
});
Loads the configuration from a toml
file. In order to use this adapter, you need to install smol-toml
(peer dependency), if you don't have it already.
npm install smol-toml
import { z } from 'zod';
import { loadConfig } from 'zod-config';
import { tomlAdapter } from 'zod-config/toml-adapter';
import path from 'path';
const schemaConfig = z.object({
MY_APP_PORT: z.string().regex(/^\d+$/),
MY_APP_HOST: z.string(),
});
const filePath = path.join(__dirname, 'config.toml');
const config = await loadConfig({
schema: schemaConfig,
adapters: tomlAdapter({ path: filePath }),
});
// using filter regex to match only the keys we need
const customConfig = await loadConfig({
schema: schemaConfig,
adapters: tomlAdapter({
path: filePath,
regex: /^MY_APP_/,
}),
});
Loads the configuration from a .env
file. In order to use this adapter, you need to install dotenv
(peer dependency), if you don't have it already. To support nested objects, you can use the nestingSeparator
property that will be used to create nested objects from flat keys based on the separator.
npm install dotenv
import { z } from 'zod';
import { loadConfig } from 'zod-config';
import { dotEnvAdapter } from 'zod-config/dotenv-adapter';
import path from 'path';
const schemaConfig = z.object({
MY_APP_PORT: z.string().regex(/^\d+$/),
MY_APP_HOST: z.string(),
});
const filePath = path.join(__dirname, '.env');
const config = await loadConfig({
schema: schemaConfig,
adapters: dotEnvAdapter({ path: filePath }),
});
// using filter regex to match only the keys we need
const customConfig = await loadConfig({
schema: schemaConfig,
adapters: dotEnvAdapter({
path: filePath,
regex: /^MY_APP_/,
}),
});
// using nesting separator to create nested objects
// .env file content: DATABASE_HOST=localhost\nDATABASE_PORT=5432
const nestedConfig = await loadConfig({
schema: z.object({
DATABASE: z.object({
HOST: z.string(),
PORT: z.string(),
}),
}),
adapters: dotEnvAdapter({
path: filePath,
nestingSeparator: '_',
}),
});
Loads configuration from TypeScript (.ts
), JavaScript (.js
), or JSON (.json
) files. The .ts
and .js
files must export a default object with the configuration data.
import { z } from 'zod';
import { loadConfig } from 'zod-config';
import { scriptAdapter } from 'zod-config/script-adapter';
import path from 'path';
const schemaConfig = z.object({
port: z.string().regex(/^\d+$/),
host: z.string(),
});
// config.ts might contain: export default { port: '3000', host: 'localhost' }
const filePath = path.join(__dirname, 'config.ts');
const config = await loadConfig({
schema: schemaConfig,
adapters: scriptAdapter({
path: filePath,
}),
});
Loads configuration from a directory containing multiple configuration files (usually used in combination with the scriptAdapter
and/or other file related adapter). Inspired by node-config, the files in the config directory are loaded in the following order:
default.EXT
default-{instance}.EXT
{deployment}.EXT
{deployment}-{instance}.EXT
{short_hostname}.EXT
{short_hostname}-{instance}.EXT
{short_hostname}-{deployment}.EXT
{short_hostname}-{deployment}-{instance}.EXT
{hostname}.EXT
{hostname}-{instance}.EXT
{hostname}-{deployment}.EXT
{hostname}-{deployment}-{instance}.EXT
local.EXT
local-{instance}.EXT
local-{deployment}.EXT
local-{deployment}-{instance}.EXT
Where EXT
is the file extension (e.g., ts
, js
, json
), instance
is the NODE_APP_INSTANCE
environment variable, deployment
is the NODE_CONFIG_ENV
or NODE_ENV
environment variables, hostname
is the HOST
, HOSTNAME
environment variables or os.hostname()
and short_hostname
is the first part of the hostname.
This adapter can be useful when using version control to manage different configurations for different environments.
import { z } from 'zod';
import { loadConfig } from 'zod-config';
import { directoryAdapter } from 'zod-config/directory-adapter';
import { scriptAdapter } from 'zod-config/script-adapter';
import path from 'path';
const schemaConfig = z.object({
port: z.string().regex(/^\d+$/),
host: z.string(),
});
const directories = [path.join(__dirname, 'config-dir')];
const config = await loadConfig({
schema,
adapters: directoryAdapter({
paths: directories,
adapters: [
{
// Restrict adapter to handle only ts files
extensions: [".ts"],
// Use the scriptAdapter for handling .ts files
adapterFactory: (filePath: string) =>
scriptAdapter({
path: filePath,
}),
},
// {
// Add here other adapters for other file types if needed
// }
],
}),
});
You can combine multiple adapters to load the configuration from different sources. We deeply merge the configuration from different sources, following the order of the adapters provided.
⚠️ Warning: When combining multiple adapters,null
values from subsequent adapters will override existing values. This behavior can be used intentionally to reset configuration values.
import { z } from 'zod';
import { loadConfig } from 'zod-config';
import { envAdapter } from 'zod-config/env-adapter';
import { jsonAdapter } from 'zod-config/json-adapter';
import path from 'path';
const schemaConfig = z.object({
port: z.string().regex(/^\d+$/),
host: z.string(),
});
const filePath = path.join(__dirname, 'config.json');
const config = await loadConfig({
schema: schemaConfig,
adapters: [
jsonAdapter({ path: filePath }),
envAdapter(),
],
});
The loadConfig
function is asynchronous to allow for adapters that are asynchronous. If you are only using synchronous adapters, you can use the loadConfigSync
function which is synchronous and does not return a promise.
The following default adapters are synchronous and can be used with loadConfigSync
:
envAdapter
jsonAdapter
json5Adapter
yamlAdapter
tomlAdapter
dotEnvAdapter
When implementing a custom adapter that you want to use with loadConfigSync
, make sure to implement the SyncAdapter
interface instead of the Adapter
interface.
import { z } from 'zod';
import { loadConfigSync } from 'zod-config';
import { envAdapter } from 'zod-config/env-adapter';
import { jsonAdapter } from 'zod-config/json-adapter';
import path from 'path';
const schemaConfig = z.object({
port: z.string().regex(/^\d+$/),
host: z.string(),
});
const filePath = path.join(__dirname, 'config.json');
const config = loadConfigSync({
schema: schemaConfig,
adapters: [
jsonAdapter({ path: filePath }),
envAdapter(),
],
});
You can use the callbacks to handle errors and success due to the async nature of the adapters.
import { z } from 'zod';
import { loadConfig } from 'zod-config';
loadConfig({
schema: schemaConfig,
onError: (error) => {
console.error('An error occurred while loading the configuration:', error);
},
onSuccess: (config) => {
console.log('Configuration loaded successfully:', config);
},
});
You can provide a custom logger to be used to log messages. By default, it uses console
.
import { z } from 'zod';
import { loadConfig, Logger } from 'zod-config';
const schemaConfig = z.object({
port: z.string().regex(/^\d+$/),
host: z.string(),
});
const customLogger: Logger = {
warn: (message) => {
// your custom implementation, e.g., log to a file or call an external service
},
};
const config = await loadConfig({
schema: schemaConfig,
logger: customLogger,
});
If any adapter fails, we will still return the data from other adapters. However, by default, we log a warning internally if an adapter fails. You can use the silent
flag to avoid logging the warning. This is useful when you have multiple adapters and you don't want to log a warning for each adapter that fails. Example for the built-in envAdapter
:
import { z } from 'zod';
import { loadConfig } from 'zod-config';
import { envAdapter } from 'zod-config/env-adapter';
const schemaConfig = z.object({
port: z.string().regex(/^\d+$/),
host: z.string(),
});
const config = await loadConfig({
schema: schemaConfig,
// silent: true --> also available in the global config in case you want to use it for all adapters
adapters: envAdapter({ silent: true }),
});
If the source of your adapters uses a different casing or formatting compared to the schema you are using, you can enable the key matching lenient
option. This is useful when working with environment variables that typically use UPPER_SNAKE_CASE or when integrating with different systems that use varying naming conventions. By default, the key matching is strict
, meaning that the keys must match exactly.
import { z } from 'zod';
import { loadConfig } from 'zod-config';
import { envAdapter } from 'zod-config/env-adapter';
const schemaConfig = z.object({
myHost: z.string(),
});
const config = await loadConfig({
schema: schemaConfig,
keyMatching: 'lenient',
adapters: envAdapter({
// keyMatching: 'lenient' --> it can also be applied to the adapter level if you want to use a different key matching for a specific adapter
}),
});
In this example, the key MYHOST
, MY_HOST
, or my-host
from the adapter would be correctly matched to myHost
in your schema.
The lenient matching works by comparing keys after:
- Removing all non-alphanumeric characters (like underscores, hyphens, dots)
- Converting to lowercase
The transform
property allows you to modify key-value pairs before they are processed by the schema. This is useful for normalizing data, filtering out unwanted keys, or transforming values. The transform function receives an object with key
and value
properties and can return either a transformed object or false
to drop the key-value pair. The transform function can be applied at both the global level (affecting all adapters) and the adapter level (affecting only that specific adapter). When both are provided, the adapter-level transform takes precedence.
Note: The transform function is the first step in the data processing pipeline, before all the other capabilities of the library (e.g., key matching, nesting separator, etc.).
import { z } from 'zod';
import { loadConfig } from 'zod-config';
import { envAdapter } from 'zod-config/env-adapter';
const schema = z.object({
database: z.object({
host: z.string(),
port: z.string(),
}),
apiKey: z.string(),
});
// Global transform - applied to all adapters
const config = await loadConfig({
schema,
transform: ({ key, value }) => {
// Drop sensitive keys
if (key.includes('SECRET')) {
return false;
}
// Transform keys to lowercase
return {
key: key.toLowerCase(),
value,
};
},
adapters: envAdapter({
customEnv: {
'DATABASE_HOST': 'localhost',
'DATABASE_PORT': '5432',
'API_KEY': 'my-key',
'SECRET_TOKEN': 'should-be-dropped',
},
nestingSeparator: '_',
}),
});
// Adapter-level transform - applied only to this adapter
const configWithAdapterTransform = await loadConfig({
schema,
adapters: envAdapter({
customEnv: {
'MY_APP_DATABASE_HOST': 'localhost',
'MY_APP_DATABASE_PORT': '5432',
'MY_APP_API_KEY': 'my-key',
'OTHER_VAR': 'ignored',
},
transform: ({ key, value }) => {
// Only process keys that start with 'MY_APP_'
if (!key.startsWith('MY_APP_')) {
return false;
}
// Remove the prefix and convert to lowercase
const cleanKey = key.replace(/^MY_APP_/, '').toLowerCase();
return {
key: cleanKey,
value,
};
},
nestingSeparator: '_',
}),
});
// Complex transformation with multiple operations
const complexConfig = await loadConfig({
schema: z.object({
api: z.object({
key: z.string(),
timeout: z.string(),
}),
database: z.object({
host: z.string(),
port: z.string(),
}),
}),
adapters: envAdapter({
customEnv: {
'API_KEY': 'secret123',
'API_TIMEOUT': '30000',
'DB_HOST': 'localhost',
'DB_PORT': '5432',
'CACHE_TTL': '3600',
},
transform: ({ key, value }) => {
// Transform different prefixes to nested structure
if (key.startsWith('API_')) {
return {
key: key.replace('API_', 'api.').toLowerCase(),
value,
};
}
if (key.startsWith('DB_')) {
return {
key: key.replace('DB_', 'database.').toLowerCase(),
value,
};
}
// Drop other keys
return false;
},
nestingSeparator: '.',
}),
});
The goal is to have a helper to load configuration data from several srouces. If you have any source in mind, feel free to open a PR to add it or just open an issue to discuss it. More adapters are coming soon.
-
This library is referenced in the official Zod documentation as part of the ecosystem section "Powered by Zod", check it out here;
-
This library is referenced in the Next.js Weekly newsletter (Packages / Tools / Repos section);
-
Check related dev.to article here;