High-Performance RESTful Java web and microservice framework.
Branch: master
Clone or download
Type Name Latest commit message Commit time
Failed to load latest commit information.
.idea Update for Intellij. Dec 28, 2018
conf Reduce logging. Dec 5, 2018
src Final 3.7 release. Last before modularization Jan 9, 2019
.gitignore Update for Intellij. Dec 28, 2018
LICENSE Update LICENSE Nov 21, 2018
README.md Update README.md Jan 8, 2019
pom.xml Final 3.7 release. Last before modularization Jan 9, 2019
proteus.iml Cleaner logging. Bake in blocking handler. Dec 28, 2018


Alt logo

An extremely lightweight, flexible, and high performance Undertow based Java framework for developing RESTful web applications and microservices.

  • Lightweight: limited dependencies and < 400kb
  • JAX-RS compliant
  • Easy on the developer and the metal
  • Blazing fast!!! The latest Techempower benchmarks demonstrate proteus outperforming 99% of all other web frameworks.

Top 5 in Java Frameworks for Fortunes

Top 5 in Java Frameworks for JSON


  • Proteus rewrites your controller methods into high performance Undertow response handlers at run time.
  • The goal is to provide the absolute highest performance while providing a simple and familiar API.
  • As developers, we feel a web framework should provide the essentials at minimal cost.

Getting Started

Quick Start

  • Make sure you have a JDK >= 8 and a current version of Maven installed.
  • Copy and paste into your terminal:
/bin/bash -e <(curl -fsSL https://raw.githubusercontent.com/noboomu/proteus-example/master/scripts/quickStart.sh)

As a dependency



Supported Controller Annotations

Controller classes respect standard Swagger / JAX-RS annotations:

@Tags({@Tag(name = "benchmarks")})

Supported Method Annotations

Controller class methods respect standard Swagger / JAX-RS annotations:

@Operation(description = "Plaintext endpoint" )
public ServerResponse<ByteBuffer> plaintext(ServerRequest request)
	return response("Hello, World!").textPlain();

Proteus has three built in annotations:

  • @Blocking

    • io.sinistral.proteus.annotations.Blocking
    • Forces the request processing to block.
  • @Debug

    • io.sinistral.proteus.annotations.Debug
    • Dumps the request and response details to the log.
  • @Chain

    • io.sinistral.proteus.annotations.Chain
    • Wraps the endpoint handler in the provided array of io.undertow.server.HttpHandler classes.

Controller methods arguments support the following JAX-RS annotations:

  • @PathParam

    • javax.ws.rs.PathParam
    • Binds a url template parameter to the method parameter.
    • i.e. if the path is /dogs/{id}, @PathParam("id") binds the path segment value to the method parameter.
  • @QueryParam

    • javax.ws.rs.QueryParam
    • Binds a HTTP query parameter to the method parameter.
  • @FormParam

    • javax.ws.rs.FormParam
    • Binds the form parameter within a request body to the method parameter.
  • @HeaderParam

    • javax.ws.rs.HeaderParam
    • Binds the value of a HTTP header to the method parameter.
  • @CookieParam

    • javax.ws.rs.CookieParam
    • Binds the value of a HTTP cookie to the method parameter.
  • @BeanParam

    • javax.ws.rs.BeanParam
    • Binds and attempts to convert the request body to an instance of the method parameter.
  • @DefaultParam

    • javax.ws.rs.DefaultParam
    • Sets the default value of a method parameter.

Return Types


For total control and maximum performance the raw HttpServerExchange can be passed to the controller method.

Methods that take an HttpServerExchange as an argument should not return a value.

In this case the method takes on full responsibility for completing the exchange.


The static method io.sinistral.proteus.server.ServerResponse.response helps create ServerResponse<T> instances that are the preferred return type for controller methods.

If the response object's contentType is not explicitly set, the @Produces annotation is used in combination with the Accept headers to determine the Content-Type.

For methods that should return a String or ByteBuffer to the client users can create responses like this:

@Operation(description = "Serve a plaintext message using a ServerResponse")
public ServerResponse<ByteBuffer> plaintext(ServerRequest request, @QueryParam("message") String message)
   return ServerResponse.response("Hello, World!").textPlain();

By default, passing a String to the static ServerResponse.response helper function will convert it into a ByteBuffer.

For other types of responses the following demonstrates the preferred style:

@Operation(description = "Return a world JSON object")
public ServerResponse<World> getWorld(ServerRequest request, @QueryParam("id") Integer id,  @QueryParam("randomNumber") Integer randomNumber )
	return response(new World(id,randomNumber)).applicationJson();

The entity can be set separately as well:

this disables static type checking!

@Operation(description = "Return a world JSON object")
public io.sinistral.proteus.server.ServerResponse getWorld(Integer id,  Integer randomNumber )
	return io.sinistral.proteus.server.ServerResponse.response().entity(new World(id,randomNumber));

CompletableFuture<ServerResponse<T>> can also be used as a response type:

@Operation(description = "Future user endpoint"  )
public CompletableFuture<ServerResponse<User>> futureUser( ServerRequest request )
	return CompletableFuture.completedFuture(response( new User(123L) ).applicationJson() );

In this case a handler will be generated with the following source code:

public void handleRequest(final io.undertow.server.HttpServerExchange exchange) throws java.lang.Exception { 
    CompletableFuture<ServerResponse<User>> response = examplesController.futureUser();
    response.thenAccept( r ->  r.applicationJson().send(this,exchange) )
    	.exceptionally( ex ->  {
      		throw new java.util.concurrent.CompletionException(ex);
  	} );

Controller Parameters

A io.sinistral.proteus.server.ServerRequest can be added as an endpoint parameter if the user wishes to access request properties that are not included in the parameter list.

Proteus is capable of parsing most types of endpoint parameters automatically so long as the type has a fromString, valueOf, or can be deserialized from JSON.

Multipart/Form file uploads can be passed to the endpoint methods as a java.io.File, a java.nio.Files.Path, or a java.nio.ByteBuffer.

Optional parameters are also supported, here is a more complex endpoint demonstrating several parameter types:

@Operation(description = "Complex parameters")
public ServerResponse<Map<String,Object>> complexParameters(
        ServerRequest serverRequest, 
        @PathParam("pathLong") final Long pathLong, 
        @QueryParam("optionalQueryString") Optional<String> optionalQueryString, 
        @QueryParam("optionalQueryLong") Optional<Long> optionalQueryLong, 
        @QueryParam("optionalQueryDate") Optional<OffsetDateTime> optionalQueryDate, 
        @QueryParam("optionalQueryUUID") Optional<UUID> optionalQueryUUID, 
        @HeaderParam("optionalHeaderUUID") Optional<UUID> optionalHeaderUUID,
        @QueryParam("optionalQueryEnum") Optional<User.UserType> optionalQueryEnum,
        @HeaderParam("optionalHeaderString") Optional<String> optionalHeaderString,
        @QueryParam("queryUUID") UUID queryUUID,  
        @HeaderParam("headerString") String headerString,
        @QueryParam("queryEnum") User.UserType queryEnum, 
	    @QueryParam("queryIntegerList") List<Integer> queryIntegerList, 
	    @QueryParam("queryLong") Long queryLong 
		Map<String,Object> responseMap = new HashMap<>(); 
		responseMap.put("optionalQueryString", optionalQueryString.orElse(null));
		responseMap.put("optionalQueryLong", optionalQueryLong.orElse(null));
	 	responseMap.put("optionalQueryDate", optionalQueryDate.map(OffsetDateTime::toString).orElse(null));
		responseMap.put("optionalQueryUUID", optionalQueryUUID.map(UUID::toString).orElse(null));
		responseMap.put("optionalHeaderUUID", optionalHeaderUUID.map(UUID::toString).orElse(null));
		responseMap.put("optionalHeaderString", optionalHeaderString.orElse(null));
		responseMap.put("optionalQueryEnum", optionalQueryEnum.orElse(null));
		responseMap.put("queryEnum", queryEnum);
		responseMap.put("queryUUID", queryUUID.toString()); 
		responseMap.put("queryLong", queryLong);
		responseMap.put("pathLong", pathLong);
		responseMap.put("headerString", headerString); 
		responseMap.put("queryIntegerList", queryIntegerList); 
		return response(responseMap).applicationJson(); 


Proteus comes with three standard services that extend the io.sinistral.proteus.services.BaseService class.

  • AssetsService

    The AssetsService mounts an asset directory at a given path and is configured in your application.conf file.

    The default configuration:

     assets {
        # the base path assets will be server from
       path = "/public"
       # the directory to load the assets from
    	 dir = "./assets"
    	 cache {
    	 # cache timeout for the assets
    	     time = 500
  • SwaggerService

    The SwaggerService generates a swagger-spec file from your endpoints and serves a swagger-ui and spec.

    The default configuration serves the spec at {application.path}/swagger.json and the ui at ${application.path}/swagger.

    The service is configured in your application.conf file.

    The default configuration:

     swagger {
    	 # the path that has an index.html template and theme css files
    	 # swagger version
     	swagger: "2.0"
    	 info {
    	     # swagger info title
    	     title = ${application.name}
    	     # swagger info version
     	    version = ${application.version}
       # swagger-ui theme from ostranmes swagger-ui-themes, the following are built-in 
       # [feeling-blue, flattop, material, monokai, muted, newspaper, outline]
       # specifying a different name causes the SwaggerService to search in 
       # {swagger.resourcePrefix}/themes for a file named "theme-{swagger.theme}.css"
       # where the swagger endpoints will be mounted
       basePath= ${application.path}"/swagger"
       #the name of the spec file
    	 consumes = ["application/json"]
    	 produces = ["application/json"]
    	 # supported schemes
    	 schemes = ["http"]
  • OpenAPIService

    The OpenAPIService generates an openapi-spec file from your endpoints and serves a swagger-ui and spec.

    The default configuration serves the spec at {application.path}/openapi.yaml and the ui at ${application.path}/openapi.

    The service is configured in your application.conf file.

    The default configuration:

     openapi {
      basePath= ${application.path}"/openapi"
      port = ${application.ports.http}
      # openapi info
    info {
     title = ${application.name}
     version = ${application.version}
     description="Proteus Server"
     securitySchemes {
    	 ApiKeyAuth = {
     servers = [
    	   description="Default Server"  

Plugins / Modules

Where are all of the plugins for everything?!?!

Proteus's design philosophy is a minimal one, so from our perspective managing a long list of plug and play plugins does not make that much sense.

Our experience with other frameworks has been that plugins are swiftly out-dated or hide too much of the underlying implementation to be useful.

However, making your own "plugin" is simple and much more gratifying.

Here is an example AWSModule that provides AWS S3 and SES support.

This example assumes you have defined the relevant aws properties in your config file:

public class AWSModule extends AbstractModule
	private static Logger log = LoggerFactory.getLogger(AWSModule.class.getCanonicalName());

	protected String accessKey;
	protected String secretKey;

	public void configure()

		AWSCredentials credentials = new BasicAWSCredentials(accessKey,  secretKey);
		AWSStaticCredentialsProvider credentialsProvider = new AWSStaticCredentialsProvider(credentials);

		AmazonS3Client s3Client = (AmazonS3Client) AmazonS3ClientBuilder.standard().withCredentials(credentialsProvider).withRegion("us-west-2").build();
		TransferManager transferManager = TransferManagerBuilder.standard().withMultipartUploadThreshold(8000000L).withS3Client(s3Client).withExecutorFactory(new ExecutorFactory()

			public ExecutorService newExecutor()
				ThreadFactory threadFactory = new ThreadFactory()
					private int threadCount = 1;

					public Thread newThread(Runnable r)
						Thread thread = new Thread(r);
						thread.setName("s3-transfer-manager-worker-" + threadCount++);
						return thread;
				return (ThreadPoolExecutor) Executors.newFixedThreadPool(Runtime.getRuntime().availableProcessors() * 2, threadFactory);

		AmazonSimpleEmailServiceAsyncClient sesClient = (AmazonSimpleEmailServiceAsyncClient) AmazonSimpleEmailServiceAsyncClientBuilder.standard().withCredentials(credentialsProvider).withRegion("us-west-2").build();

Now you can simply inject these in any controller by adding the following lines to the top of the controller:

	protected TransferManager transferManager;
	protected AmazonS3Client s3Client;
	protected  AmazonSimpleEmailServiceAsyncClient sesClient;

Also, please note that the implementation of ProteusApplication you are using would also need to add the following line before starting:


Under the Hood

Proteus takes your MVC controller classes and methods decorated with Swagger / JAX-RS annotations and generates native Undertow handler classes at runtime.

By default, the configuration is loaded into a com.typesafe.config.Config from a file at conf/application.conf.

@Named annotated properties of Module, Service and controller classes are bound to values found in the configuration.

Proteus applications generally have a main method that creates an instance of io.sinistral.proteus.ProteusApplication.

Prior to calling start on the ProteusApplication instance:

  • Register Service classes via registerService
  • Register Module classes via registerModule
  • Register classes annotated with javax.ws.rs.Path via registerController

Out of the box you get a Swagger UI at /openapi.

A Service extends com.google.common.util.concurrent.AbstractIdleService or io.sinistral.proteus.services.BaseService.

A Module implements com.google.inject.Module.


Check out this example that also demonstrates pac4j integration.


  • Several years of working with the Play framework convinced us there had to be a better way.
  • We faced a Goldilocks Crisis with the existing alternatives: Jooby did too much, light-4j didn't do quite enough.
  • We needed a framework that enabled us to write clean MVC REST controllers that created Swagger docs we could plug directly into the existing codegen solutions.
  • We needed a framework with minimal overhead and performance at or near that of raw Undertow.

Inspired by Play, Jooby, and light-4j.


Built With