Building your own plugin is easy, and if you're already familiar with Payload then you'll have everything you need to get started. You can either start from scratch or use the Payload plugin template to get up and running quickly.
Our plugin template includes everything you need to build a full life-cycle plugin:
By abstracting your code into a plugin, you'll be able to reuse your feature across multiple projects and make it available for other developers to use.
Here is a brief recap of how to integrate plugins with Payload, to learn more head back to the plugin overview page.
To install any plugin, simply add it to your Payload config in the plugins array.
The initialization process goes in the following order:
In the Payload plugin template, you will see a common file structure that is used across plugins:
In the root folder, you will see various files related to the configuration of the plugin. We set up our environment in a similar manner in Payload core and across other projects. The only two files you need to modify are:
The purpose of the dev folder is to provide a sanitized local Payload project. so you can run and test your plugin while you are actively developing it.
Do not store any of the plugin functionality in this folder - it is purely an environment to assist you with developing the plugin.
If you're starting from scratch, you can easily setup a dev environment like this:
If you're using the plugin template, the dev folder is built out for you and the
samplePlugin has already been installed in
You can add to the
dev/payload.config and build out the dev project as needed to test your plugin.
When you're ready to start development, navigate into this folder with
And then start the project with
yarn dev and pull up
http://localhost:3000 in your browser.
Another benefit of the dev folder is that you have the perfect environment established for testing.
Jest organizes tests into test suites and cases. We recommend creating tests based on the expected behavior of your plugin from start to finish. Read more about tests in the Jest documentation.
The plugin template provides a stubbed out test suite at
dev/plugin.spec.ts which is ready to go - just add in your own test conditions and you're all set!
For development and testing, you will likely need some data to work with. You can streamline this process by seeding and dropping your database - instead of manually entering data.
In the plugin template, you can navigate to
dev/src/server.ts and see an example seed function.
A sample seed function has been created for you at
dev/src/seed, update this file with additional data as needed.
Now that we have our environment setup and dev project ready to go - it's time to build the plugin!
First up, the
src/index.ts file - this is where the plugin should be imported from. It is best practice not to build the plugin directly in this file, instead we use this to export the plugin and types from their respective files.
To reiterate, the essence of a payload plugin is simply to extend the Payload config - and that is exactly what we are doing in this file.
configto be equal to a copy of the existing config.
We are going to use spread syntax to allow us to add data to existing arrays without losing the existing data. It is crucial to spread the existing data correctly, else this can cause adverse behavior and conflicts with Payload config and other plugins.
Let's say you want to build a plugin that adds a new collection:
First, you need to spread the
config.collections to ensure that we don't lose the existing collections. Then you can add any additional collections, just as you would in a regular payload config.
This same logic is applied to other properties like admin, globals, hooks:
Some properties will be slightly different to extend, for instance the
If you wish to add to the
onInit, you must include the async/await. We don't use spread syntax in this case, instead you must await the existing
onInit before running additional functionality.
In the template, we have stubbed out a basic
onInitExtension file that you can use, if not needed feel free to delete it.
If any of your files use server only packages such as fs, stripe, nodemailer, etc, they will need to be removed from the browser bundle. To do that, you can alias the file imports with webpack.
When files are bundled for the browser, the import paths are essentially crawled to determine what files to include in the bundle. To prevent the server only files from making it into the bundle, we can alias their import paths to a file that can be included in the browser. This will short-circuit the import path crawling and ensure browser only code is bundled.
Webpack is another part of the Payload config that can be a little more tricky to extend. To help here, the template includes a helper function
extendWebpackConfig() which takes care of spreading the existing webpack, so you can just add your new stuff:
If your plugin has options, you should define and provide types for these options in a separate file which gets exported from the main
If possible, include JSDoc comments to describe the options and their types. This allows a developer to see details about the options in their editor.
In addition to the setup covered above, here are other best practices to follow:
For a better user experience, provide a way to disable the plugin without uninstalling it. This is especially important if your plugin adds additional webpack aliases, this will allow you to still let the webpack run to prevent errors.
If you've configured tests for your package, integrate them into your workflow to run the tests each time you commit to the plugin repository. Learn more about how to configure tests into your GitHub CI workflow.
The best way to share and allow others to use your plugin once it is complete is to publish an NPM package. This process is straightforward and well documented, find out more about creating and publishing a NPM package here.
Apply the tag payload-plugin to your GitHub repository. This will boost the visibility of your plugin and ensure it gets listed with existing payload plugins.
With the SemVar system you release version numbers that reflect the nature of changes (major, minor, patch). Ensure all major versions reference their Payload compatibility.