This is a plugin makes it possible to keep your function and resource definitions in separate files in a directory structure instead of the serverless.yml

#Installation External Plugins are added on a per service basis and are not applied globally. Make sure you are in your Service's root directory, then install the corresponding Plugin with the help of NPM:

npm install --save serverless-dir-config-plugin

In your service create a directory structure like

mkdir serverless
mkdir serverless/functions
mkdir serverless/resources

The plugin will look in these directories for files ending in .yml to load in. You can nest them any way you want. They use the same structure of yaml as the way they are included in the serverless.yml. They also support all the variable resolution that the main file supports.

hello:
 handler: lib/handler.hello
 role: BasicLambdaRole
 events:
 - http:
 path: hello
 method: get
 integration: lambda-proxy

Keep in mind for the resources there are multiple resource types. I tend to create directories for different groupings (ima, dynamodb,etc)

Make sure you include the type first (Resources/Outputs etc):

Resources:
 BasicLambdaRole:
 Type: AWS::IAM::Role
 Properties:
 RoleName: ${self:service}-${self:provider.stage}-BasicLambdaRole
 AssumeRolePolicyDocument:
 Version: '2012-10-17'
 Statement:
 - Effect: Allow
 Principal:
 Service:
 - lambda.amazonaws.com
 Action: sts:AssumeRole
 Policies:
 - PolicyName: ${self:service}-${self:provider.stage}-BasicLambdaPolicy
 PolicyDocument:
 Version: '2012-10-17'
 Statement:
 - Effect: Allow
 Action:
 - "logs:*"
 - "cloudwatch:*"
 Resource: "*"

You need to add it to the plugin section of the serverless.yml

plugins:
 - serverless-dir-config-plugin

Currently the only thing you can configure is if it shows what resources and functions are being loaded. If you don't want to see that - set quiet to true.

custom:
 dirconfig:
 quiet: false

Serverless.js doesn't provide hooks to append to the config file. This plugin ends up making some assumptions to work.

1. Currently the framework loads all the plugins before it does anything. This allows the plugin to look for the files and append them to the configuration before anything happens.

2. After the configuration is loaded all the variables get resolved. The plugin monkeypatches this process so it can add in function names once the variables are all set.

This means that right now the way the plugin works is very fragile. If the Serverless.js changes the way they load things - this plugin will break. I am hoping that after confirming that this plugin is useful I will submit a PR that looks at giving plugins access to the hooks that I'm basically inventing.