@artinet/sdk 0.5.4 → 0.5.6

package/README.md

@@ -7,21 +7,22 @@
 
 # artinet SDK
 
-The artinet SDK is a Full Stack Agent builder written in TypeScript for [node.js](https://nodejs.org/) that aims to simplify the creation of interoperable AI agents. Learn more at [the artinet project](https://artinet.io/).
+The artinet SDK is a TypeScript library designed for Agentic Communication. It's written for [node.js](https://nodejs.org/) and aims to simplify the creation of interoperable AI agents. Learn more at [the artinet project](https://artinet.io/).
 
-This SDK significantly supports multiple agentic communication protocols, offering a production-ready solution with a focus on developer experience, reliability, and comprehensive features.
+This SDK leverages a service-oriented architecture for building AI agents allowing developers to easily create agents as simple processes or seamlessly embed them within a dedicated server.
 
 ### Quick Start
 
-To build your own agent/server use the [`create-quick-agent`](https://www.npmjs.com/package/@artinet/create-quick-agent) command:
+To build your own agent/server use the [`create-quick-agent`](https://www.npmjs.com/package/@artinet/create-quick-agent) (supports v0.5.2) command:
 
 ```bash
 npx @artinet/create-quick-agent@latest
 ```
 
-Its got [serveral template projects](https://github.com/the-artinet-project/create-quick-agent) that you can use to get started building agents today.
+It has [serveral template projects](https://github.com/the-artinet-project/create-quick-agent) that you can use to get started building agents today.
 
 ## Table of Contents
+
 - [artinet SDK](#artinet-sdk)
     - [Quick Start](#quick-start)
   - [Table of Contents](#table-of-contents)
@@ -39,33 +40,39 @@ Its got [serveral template projects](https://github.com/the-artinet-project/crea
       - [Authentication](#authentication)
     - [Server](#server)
       - [Implementing an A2A Agent](#implementing-an-a2a-agent)
+        - [AgentBuilder](#agentbuilder)
+        - [AgentEngine](#agentengine)
+      - [Event Handling/Monitoring \& Message Streaming](#event-handlingmonitoring--message-streaming)
       - [Persistent Storage](#persistent-storage)
       - [Logging](#logging)
-      - [Server Registration \& Discovery](#server-registration--discovery)
+      - [Server Registration \& Discovery (v0.5.2)](#server-registration--discovery-v052)
       - [Advanced Server Customization](#advanced-server-customization)
-    - [Quick-Agents (Alpha)](#quick-agents-alpha)
+    - [Quick-Agents (Alpha v0.5.2)](#quick-agents-alpha-v052)
   - [Contributing](#contributing)
   - [License](#license)
   - [Acknowledgements](#acknowledgements)
 
 ## Features
 
-- **Plug-and-Play Server:** Built on Express.js, the `ExpressServer` handles Transport-layer complexity, routing, protocol compliance, and Server-Sent Events (SSE) streaming mechanics automatically. Just provide your core agent logic (`AgentEngine`) and configuration via `ExpressServerOptions`.
-- **TypeScript First:** Fully written in TypeScript with comprehensive type definitions for a robust developer experience.
-- **Protocol Compliance:** Implements the complete A2A specification with growing support for recieving commands via MCP & ACP.
-- **Code Deployment (Experimental)** | Bundle, test, and deploy agent code onto the artinet. Includes bundler, task wrapper, and deployment utilities.
-
-| Component/Feature   | Description                                                                 | Key Classes/Types                                                                            |
-| :------------------ | :-------------------------------------------------------------------------- | :------------------------------------------------------------------------------------------- |
-| **Client**          | Interact with A2A-compliant agents. Supports standard & streaming requests. | `A2AClient`, `RpcError`                                                                      |
-| **Server**          | Host agents that can communicate using a variety of protocols. Handles protocol details, routing, SSE.          | `ExpressServer`, `ExpressServerOptions`                                                               |
-| **Reqeust Handling**   | Define agent logic using async generators.                                  | `AgentEngine`, `ExecutionContext`, `UpdateEvent`                                              |
-| **Storage**         | Persist event state. In-memory and file-based options included.              | `Store`, `InMemoryTaskStore`, `FileStore`                                                |
-| **Streaming (SSE)** | Handle real-time updates via SSE.   |                                           |
-| **Logging**         | Configure structured logging for debugging and monitoring.                  | `logger`, `configureLogger`, `LogLevel`                                                      |
-| **Advanced Server** | Customize the underlying JSON-RPC server or integrate into existing apps.   | `CreateJSONRPCServerParams`, `createJSONRPCMethod` |
-| **Core Types**      | Based on the official JSON Schemas.                                      | `Tool`,`AgentCard`, `Task`, `Message`, `Part`, `Artifact`, etc.                                     |
-| **Agent Utilities (for Sandboxed Environments)** | Standardized utilities for agents in managed environments to interact with the host system for task lifecycle, inter-agent communication, and external API calls. | `artinet.v0.taskManager`, `artinet.v0.connect`, `artinet.v0.agent`, `TaskProxy`, `ConnectAPICallback`, `ClientProxy`, `ClientFactory` |
+- **Service-Oriented Architecture:** Modern, modular design with a dedicated layer for protocol handling and state management.
+- **Modern Express Server:** Quickly create an Express Server with the `createAgentServer()` function. It handles all of the transport-layer complexity, adds an A2A <-> JSON-RPC middleware, and manages Server-Sent Events (SSE) automatically.
+- **TypeScript First:** Written in TypeScript with a fully implemented Zod <-> A2A schema.
+- **Protocol Compliance:** Implements the complete A2A specification (v0.3.0) with support for any kind of transport layer (Express, tRPC, WebSockets, etc).
+- **Code Deployment:** Bundle, test, and deploy agent code onto the artinet via the `./deployment` module (v0.5.3). Includes bundler, agent handler, and deployment utilities.
+
+| Component/Feature    | Description                                                                 | Key Classes/Types                                                                                    |
+| :------------------- | :-------------------------------------------------------------------------- | :--------------------------------------------------------------------------------------------------- |
+| **Client**           | Interact with A2A-compliant agents. Supports standard & streaming requests. | `A2AClient`                                                                                          |
+| **Server**           | Quickly spin up an Agent as a Server.                                       | `createAgentServer`, `ExpressAgentServer`, `jsonRPCMiddleware`                                       |
+| **Agent Creation**   | Design agents with the Agent Builder or take control with the Agent Engine. | `AgentEngine`, `Context`, `AgentBuilder`                                                             |
+| **Storage**          | Persist event state. In-memory and file-based options included.             | `Store`, `InMemoryTaskStore`, `FileStore`                                                            |
+| **Streaming (SSE)**  | Handle real-time updates via Server-Sent Events with middleware support.    | Built-in SSE handling, streaming generators                                                          |
+| **Event Handling**   | Monitor agent execution with our event system.                              | `EventManager`, `eventOverrides`, Event listeners (`start`, `update`, `error`, `complete`, `cancel`) |
+| **Logging**          | Configure structured logging for debugging and monitoring.                  | `logger`, `configureLogger`, `LogLevel`                                                              |
+| **Transport Layers** | Built-in support for Express/TRPC.                                          | Express middleware, tRPC router, or easily create your own.                                          |
+| **Core Types**       | Fully implement A2A schema in Zod.                                          | `AgentCard`, `Task`, `Message`, `Part`, `Artifact`, etc.                                             |
+| **Deployment**       | Bundle, test, and deploy agents onto the artinet platform (v0.5.2).         | `@artinet/sdk/deployment`, `fullDeployment`, `testDeployment`, `bundle`                              |
+| **Agent Utilities**  | Run agents in managed environments with our proxy system.                   | `artinet.v0.taskManager`, `artinet.v0.connect`, `artinet.v0.agent`                                   |
 
 ## Installation
 
@@ -75,59 +82,73 @@ npm install @artinet/sdk
 
 ## Requirements
 
-- Node.js (v22.0.0 or higher recommended, check `package.json` engines for exact requirement)
+- Node.js (v22.0.0 or higher recommended)
 
 ## Documentation
 
 For more detailed documentation visit our documentation site [here](https://the-artinet-project.github.io/artinet-documentation/).
 
 ## Example
- 
-A basic A2A server and client interaction. For more detailed examples, see the `examples/` directory.
+
+A basic A2A server and client interaction (For simple agents see the [AgentBuilder](#agentbuilder) section). For more detailed examples, see the `examples/` directory.
 
 **1. Server (`quick-server.ts`)**
 
 ```typescript
 import {
-  ExpressServer,
+  createAgentServer,
   ExecutionContext,
   AgentEngine,
-  InMemoryTaskStore,
+  TaskManager,
+  MessageSendParams,
+  getParts,
 } from "@artinet/sdk";
 
-// Minimal agent logic: receive text, yield working state, yield completed state with echo
-const quickAgentLogic: AgentEngine = async function* (context: ExecutionContext) {
-  const params = context.getRequestParams() as MessageSendParams; //for A2A requests
-  const userInput = params.message.parts[0].kind === "text"
-      ? context.userMessage.parts[0].text
-      : "";
-  yield { state: "working" };
-  // Simulate some work if needed, check context.isCancelled()
+//Define your Agents Behaviour
+const quickAgentLogic: AgentEngine = async function* (context: Context) {
+  const params: MessageSendParams = context.command; //MessageSendParams
+  const { text: userInput} = getParts(params.message.parts); //helper function to consume parts
+  const task = context.State(); //TaskAndHistory
+  yield {
+      ...
+      state: "working"
+      ...
+    };
+
   yield {
+    ...
     state: "completed",
     message: {
       role: "agent",
       parts: [{ kind: "text", text: `You said: ${userInput}` }],
     },
+    ...
   };
 };
 
-const server = new A2AServer({
-  handler: quickAgentLogic,
-  taskStore: new InMemoryTaskStore(),
-  port: 4000,
-  basePath: "/a2a",
-  card: {
-    name: "QuickStart Agent",
-    url: "http://localhost:4000/a2a",
-    version: "0.1.0",
-    capabilities: { streaming: true },
-    skills: [{ id: "echo", name: "Echo Skill" }],
+// Create an agent server
+const { app, agent } = createAgentServer({
+  agent: {
+    engine: quickAgentLogic,
+    agentCard: {
+      name: "QuickStart Agent",
+      url: "http://localhost:4000/a2a",
+      version: "0.1.0",
+      capabilities: { streaming: true },
+      skills: [{ id: "echo", name: "Echo Skill" }],
+      ...
+    },
+    tasks: new TaskManager(),
+    ...
   },
+  basePath: "/a2a",
+  agentCardPath: "/.well-known/agent.json"
+  ...
 });
 
-server.start();
-console.log("A2A Server running at http://localhost:4000/a2a");
+app.listen(4000, () => {
+  console.log("A2A Server running at http://localhost:4000/a2a");
+});
 ```
 
 **2. Client (`quick-client.ts`)**
@@ -142,13 +163,14 @@ async function runClient() {
     messageId: "test-message-id",
     kind: "message",
     role: "user",
-    parts: [{ kind: "text", text: "Hello Quick Start!" }],
+    parts: [{ kind: "text", text: "Hello World!" }],
+    ...
   };
 
   const stream = client.sendStreamingMessage({ message });
 
   for await (const update of stream) {
-    // process the update
+    // process the updates
     ...
   }
   console.log("Stream finished.");
@@ -193,6 +215,7 @@ async function runBasicTask() {
     kind: "message",
     role: "user",
     parts: [{ kind: "text", text: "What is the capital of France?" }],
+    ...
   };
 
   const task = await client.sendMessage({ message });
@@ -216,10 +239,13 @@ async function runStreamingTask() {
   const client = new A2AClient("https://your-a2a-server.com/a2a");
   const message: Message = {
     role: "user",
-    parts: [{ type: "text", text: "Tell me a short story." }],
+    parts: [{ kind: "text", text: "Tell me a short story." }],
+    ...
   };
 
-  const stream = client.sendStreamingMessage({ id: "streaming-task-1", message });
+  const stream = client.sendStreamingMessage({
+    message,
+  });
 
   for await (const update of stream) {
     if ((update as TaskStatusUpdateEvent).status) {
@@ -253,76 +279,231 @@ client.setHeaders({ Authorization: "Bearer ...", "X-Custom": "value" });
 
 ### Server
 
-Host agents using `A2AServer`. Handles protocol details. See `examples/` for more.
+Use `createAgentServer()` to embed your Agents in an Express App.
 
 #### Implementing an A2A Agent
 
-Define agent behavior with an async generator `TaskHandler`.
+The SDK provides a variety of options for creating complex ([AgentEngine](#agentengine)) or simple agents ([AgentBuilder](#agentbuilder)).
+
+##### AgentBuilder
+
+**Option 1: Using AgentBuilder (Recommended for Simple Workflows)**
+
+For simple agents with one or more processing steps, use the `AgentBuilder` pattern:
+
+```typescript
+import { createAgentServer, AgentBuilder, TaskManager } from "@artinet/sdk";
+
+//create a simple agent
+const simpleAgent = AgentBuilder().text(() => "hello world!");
+
+//or design a powerful workflow
+const { app, agent } = createAgentServer({
+  agent: AgentBuilder()
+    .text(({ command, context }) => {
+      const userMessage =
+        command.message.parts[0]?.kind === "text"
+          ? command.message.parts[0].text
+          : "";
+      return {
+        parts: [`Processing: ${userMessage}`], //parts are immediately sent back to the caller as TaskStatusUpdateEvents
+        args: [userMessage], //args are passed to the next step
+      };
+    })
+    .file(({ command, args }) => {
+      const processedText = args[0];
+      return {
+        parts: [
+          {
+            name: "result.txt",
+            mimeType: "text/plain",
+            bytes: `Processed result: ${processedText}`,
+          },
+        ],
+        args: ["file-generated"],
+      };
+    })
+    .text(({ command, args }) => {
+      const status = args[0];
+      return `Task completed. Status: ${status}`;
+    }) //A final Task is returned to the caller which contains all of the emitted parts.
+    .createAgent({
+      agentCard: {
+        name: "Multi-Step Agent",
+        url: "http://localhost:3000/a2a",
+        version: "1.0.0",
+        capabilities: { streaming: true },
+        skills: [{ id: "multi-processor", name: "Multi-Step Processor" }],
+      },
+      tasks: new TaskManager(),
+    }),
+  basePath: "/a2a",
+});
+
+app.listen(3000, () => {
+  console.log("Multi-Step A2A Server running on http://localhost:3000/a2a");
+});
+```
+
+The `AgentBuilder` approach is particularly useful when you need:
+
+- **Step-by-step processing**: Break down complex tasks into discrete, manageable steps
+- **Data flow between steps**: Pass results from one step to the next using the `args` parameter
+- **Different content types**: Mix text, file, and data processing in a single workflow
+- **Reusable components**: Build modular agents that can be easily edited or extended
+
+##### AgentEngine
+
+**Option 2: Direct AgentEngine Implementation**
+
+When you need full control over the execution flow, implement an `AgentEngine` directly:
 
 ```typescript
 import {
-  A2AServer,
-  ExecutionContext,
+  createAgentServer,
+  Context,
   AgentEngine,
-  InMemoryTaskStore,
+  TaskManager,
 } from "@artinet/sdk";
 
-
-const myAgent: AgentEngine = async function* (context: ExecutionContext) {
-
+const myAgent: AgentEngine = async function* (context: Context) {
+  const task: TaskAndHistory = context.events.getState();
   yield {
     state: "working",
     message: {
+      ...
       role: "agent",
-      parts: [{ type: "text", text: "Processing..." }],
+      parts: [{ kind: "text", text: "Processing..." }],
     },
+    ...
   };
-
-  // Check context.isCancelled() if operation is long
-  // await someAsyncTask();
-  ...
-
   yield {
+    ...
     name: "result.txt",
     mimeType: "text/plain",
-    parts: [{ type: "text", text: "Report data" }],
+    parts: [{ kind: "text", text: "Report data" }],
   };
 
   yield {
+    ...
     state: "completed",
     message: {
+      kind: "message"
       role: "agent",
-      parts: [{ type: "text", text: "Finished processing." }],
+      parts: [{ kind: "text", text: "Finished processing." }],
+      ...
     },
   };
 };
 
-const myServer = new A2AServer({
-  handler: myAgent,
-  taskStore: new InMemoryTaskStore(),
-  port: 3000,
+const { app, agent } = createAgentServer({
+  agent: {
+    engine: myAgent,
+    agentCard: {
+      name: "Example Agent",
+      url: "http://localhost:3000/a2a",
+      version: "1.0.0",
+      capabilities: { streaming: true },
+      skills: [{ id: "processor", name: "Text Processor" }],
+    },
+    tasks: new TaskManager(),
+  },
   basePath: "/a2a",
-  card: {
-    name: "Example Agent",
+  agentCardPath: "/.well-known/agent-card.json",
+});
+```
+
+#### Event Handling/Monitoring & Message Streaming
+
+The SDK provides comprehensive event handling & message streaming capabilities that allow you to modify agent execution, subscribe to events, stream commands, and respond to state changes in real-time.
+
+**Override Event Behaviour**
+
+When using the service layer, you can provide your own Event Handlers:
+
+```typescript
+import { createAgent, TaskManager, ContextManager, Command, SendCommandInterface } from "@artinet/sdk";
+
+const customContextManager = new ContextManger();
+const agent = createAgent({
+  engine: (context: Context){
+    context.events.on("update", (currentState, nextState) => {
+      //allow other processes to subscribe to your agent
+    })
+    ...
+    //handle command streams directly within an agent
+    for await (const command of context.command) {
+      console.log("new command recieved: ", command);
+      //will continue polling until the command stream is closed by calling command.close();
+    }
+    //or process them asynchronously
+    context.command.on("send", (command) => {
+      ...
+    });
+  },
+  agentCard: {
+    name: "Event-Monitored Agent",
     url: "http://localhost:3000/a2a",
     version: "1.0.0",
     capabilities: { streaming: true },
-    skills: [{ id: "processor", name: "Text Processor" }],
+    skills: [{ id: "monitor", name: "Monitored Agent" }],
+  },
+  contexts: customContextManager,
+  tasks: new TaskManager(),
+  eventOverrides: { //for even greater control create your own Event Handlers
+    onStart: async (context) => {
+      ...
+      return currentState;
+    },
+    onUpdate: async (currentState, nextState) => {
+      ...
+      return currentState;
+    },
+    ...
   },
 });
 
-myServer.start();
-console.log("A2A Server running on http://localhost:3000/a2a");
+const result = await agent.sendMessage({
+  contextId: "123"
+  ...
+});
+
+const currentContext = customContextManager.getContext("123");
+
+//subscribe to the events from a specific context
+currentContext.events.on("complete", () {
+  ...
+  //errors thrown here will be triggered in the original context
+});
+
+//stream new commands into a running context
+(currentContext.command as SendCommandInterface<Command>).send({
+  ...
+});
+
+currentContext.command.close();
 ```
 
+**Available Event Types**
+
+The EventManager supports the following event types:
+
+- **`OnStart`/`start`**: Fired when agent execution begins
+- **`OnUpdate`/`update`**: Fired on each state update during execution
+- **`OnError`/`error`**: Fired when an error occurs during execution
+- **`OnComplete`/`complete`**: Fired when agent execution completes successfully
+- **`OnCancel`/`cancel`**: Fired when agent execution is cancelled
+
 #### Persistent Storage
 
-Use `FileStore` for file-based persistence. Ensure the directory exists.
+For storage, use our out of the box storage providers like `FileStore`. Or implement the `Store` interface to create your own.
 
 ```typescript
 import path from "path";
 import fs from "fs";
+import { FileStore } from "@artinet/sdk";
 
+//make sure the directory exists
 const dataDir = path.join(process.cwd(), "a2a-data");
 if (!fs.existsSync(dataDir)) {
   fs.mkdirSync(dataDir, { recursive: true });
@@ -330,8 +511,14 @@ if (!fs.existsSync(dataDir)) {
 
 const myStore = new FileStore(dataDir);
 
-const myServer = new A2AServer({
-  taskStore: myStore,
+const { app, agent } = createAgentServer({
+  agent: {
+    engine: myAgent,
+    agentCard: {
+      ...
+    },
+    tasks: myStore,
+  },
   ...
 });
 ```
@@ -350,108 +537,160 @@ logger.info("Server starting...");
 //use helper functions
 logDebug("LoggerTest", { taskId: "task-123" }, "Task status updated.");
 
-// Create child logger with bound context
+// Create child loggers with bounded contexts
 const taskLogger = logger.child({ taskId: "abc" });
 taskLogger.info("Processing step X");
 ```
 
-#### Server Registration & Discovery
+#### Server Registration & Discovery (v0.5.2)
 
-The SDK includes features to help make your agent discoverable:
+The SDK includes features to help make your agent discoverable using the new service-based architecture:
 
-- **Automatic Registration:** You can configure your `A2AServer` to automatically register your `AgentCard` with the [A2A Registry](https://artinet.io) upon startup by setting `register: true` (default: `false`) in the server parameters.
+- **Automatic Registration:** You can configure your agent server to automatically register your `AgentCard` with the [A2A Registry](https://artinet.io) upon startup by setting `register: true` in the server parameters (temporarily unavailable in v0.5.6).
 
 ```typescript
-  const myServer = new A2AServer({
-    ...
-    card: {
-      ...
+const { app, agent } = createAgentServer({
+  agent: {
+    engine: myAgent,
+    agentCard: {
+      // ...
       url: "http://my-public-domain:3000/my-agent", // Publicly accessible URL
-      ...
+      // ...
     },
-    register: true, // Enable automatic registration on start
-  });
+  },
+});
 ```
 
-- **Custom Agent Card Path:** By default, the server exposes its `AgentCard` at `/.well-known/agent.json` [RFC8615](https://datatracker.ietf.org/doc/html/rfc8615) and a fallback at `/agent-card`. You can specify a different fallback path using the `fallbackPath` option in `A2AServerParams`.
+- **Custom Agent Card Path:** By default, the server exposes its `AgentCard` at `/.well-known/agent-card.json` following [RFC8615](https://datatracker.ietf.org/doc/html/rfc8615). You can specify a custom path using the `agentCardPath` option.
 
 ```typescript
-  const myServer = new A2AServer({
-    ...
-    basePath: "/apiV2"
-    fallbackPath: "/apiV2/custom-card-info", // Agent card available here
-    ...
-  });
-// The AgentCard is now accessible at http://localhost:3001/apiV2/custom-card-info
+const { app, agent } = createAgentServer({
+  ...
+  basePath: "/apiV2",
+  agentCardPath: "/apiV2/custom-card-info", // Custom agent card path
+});
+// The AgentCard is now accessible at http://localhost:3000/apiV2/custom-card-info
 ```
 
 #### Advanced Server Customization
 
-Provide a custom `createJSONRPCServer` function (implementing `JSONRPCServerFactory`) for fine-grained control over the underlying RPC server.
+Our new architecture provides multiple ways to customize your agent server:
+
+**1. Using `createAgentServer`:**
+Easily spin up an A2A Express app `createAgentServer()`:
+
+```typescript
+const initialApp = express();
+
+// custom middleware
+initialApp.use((req, res, next) => {
+  ...
+  next();
+});
 
-This factory function receives objects of type `CreateJSONRPCServerParams` & `RequestParams` containing the necessary SDK dependencies (`taskHandler`, `taskStore`, `agentCard`, `activeCancellations`, `createTaskContext`, `closeStreamsForTask`) and the specific method parameters (i.e. `SendMessageRequest["params"]`). You can use these dependencies to configure the standard A2A methods and add your own custom JSON-RPC methods.
+const { app, agent } = createAgentServer({
+  app: initialApp
+  agent: {
+    ...
+  },
+});
 
-The SDK exports default handlers for the standard A2A methods (e.g., `defaultSendMessageMethod`), create your own using dedicated A2A method types(`SendMessageMethod`) and use `createJSONRPCMethod` to easily wrap these methods with dependency injection and error handling.
+// custom middleware
+app.use("/custom", (req, res, next) => {
+  ...
+});
 
-See `src/server/lib/middleware/factory.ts` and `src/server/lib/middleware/a2a-methods.ts` for implementation details.
+```
 
-**Example:**
+**2. Use the JSON-RPC Middleware:**
+Directly import our preconfigured JSON-RPC middleware:
 
 ```typescript
+import express from "express";
+import { createAgent, jsonRPCMiddleware, errorHandler, InMemoryTaskStore } from "@artinet/sdk";
 
-const myCustomSendMethod: SendMessageMethod = (
-  deps,
-  requestParams,
-  callback
-) => {
-  const { taskStore, taskHandler, createTaskContext } = deps;
-  const { id: taskId } = requestParams;
-  const { message, sessionId, metadata } = requestParams;
-  ...
-  callback(null, ...);
-};
+const customApp = express();
 
-const myCustomRPCServer: JSONRPCServerFactory = (
-  params: CreateJSONRPCServerParams
-): JSONRPCServerType => {
-  //Use a custom message/send method
-  const messageSendMethod = createJSONRPCMethod(params, myCustomSendMethod, "message/send");
-  const taskGetMethod = createJSONRPCMethod(params, defaultGetTaskMethod, "tasks/get");
-  const taskCancelMethod = createJSONRPCMethod(params, defaultCancelTaskMethod, "tasks/cancel");
-
-  // Note: Push notifications are not fully implemented yet
-  const taskPushNotificationSetMethod = createJSONRPCMethod(params, defaultSetTaskPushNotificationMethod, "tasks/pushNotificationConfig/set");
-  const taskPushNotificationGetMethod = createJSONRPCMethod(params, defaultGetTaskPushNotificationMethod, "tasks/pushNotificationConfig/get");
-
-  const rpcServer = new JSONRPCServer({
-    "message/send": messageSendMethod,
-    "tasks/get": taskGetMethod,
-    "tasks/cancel": taskCancelMethod,
-    "tasks/pushNotificationConfig/set": taskPushNotificationSetMethod,
-    "tasks/pushNotificationConfig/get": taskPushNotificationGetMethod,
-  });
+const agent = createAgent({
+  engine: myAgentLogic,
+  agentCard: {
+    ...
+  },
+  tasks: new InMemoryTaskStore(),
+});
 
-  return rpcServer;
-};
+customApp.use("/auth", yourAuthMiddleware);
+customApp.use("/metrics", yourMetricsMiddleware);
+customApp.use(express.json());
 
+// Add the A2A middleware
+customApp.post("/", async (req, res, next) => {
+  return await jsonRPCMiddleware(agent, req, res, next);
+});
+// Dont forget to add error handling*
+customApp.use(errorHandler);
 
-const server = new A2AServer({
-  createJSONRPCServer: myCustomRPCServer,
-  ...
+// Serve the agent card
+customApp.get("/.well-known/agent-card.json", (req, res) => {
+  res.json(agent.agentCard);
+});
+
+// Start your custom server
+const server = customApp.listen(3000, () => {
+  console.log("Custom A2A server running on port 3000");
 });
 ```
 
-**Using the Custom Factory**
+**3. Using Custom Transport Layers:**
+Use our preconfigured TRPC router, or create your own integration with WebSockets & other protocols:
 
-Pass your factory function via the `createJSONRPCServer` option during `A2AServer` initialization.
+```typescript
+import { createAgentRouter } from "@artinet/sdk";
 
-**Important:** The default `A2AServer` setup automatically adds Express middleware to handle Server-Sent Events (SSE) for `message/stream` and `tasks/resubscribe`, as well as the `/agent/card` (and `/.well-known/agent.json`) GET endpoints. If you are **not** using `A2AServer` and integrating the Jayson server middleware into your own Express application, you **must** implement these SSE and card endpoints yourself to maintain full A2A compliance, especially for streaming functionality. See `src/server/lib/express-server.ts` for how the default server handles these routes.
+const agentRouter = createAgentRouter();
+```
 
-### Quick-Agents (Alpha)
+**Use the Agent:**
+Directly invoke the agent to use it locally:
 
-We are excited to introduce new capabilities for deploying agents directly onto the artinet. 
+```typescript
+import { createAgent } from "@artinet/sdk";
 
-We've added a `testDeployment` utility which is available for all users letting you bundle and test your agent logic in a temporary sandboxed environment. 
+const agent = createAgent({
+  engine: myAgentLogic,
+  agentCard: {
+    ...
+  },
+  tasks: new InMemoryTaskStore(),
+});
+
+// Wrap these calls in your desired transport logic
+const result = await agent.sendMessage({
+  ...
+});
+
+// Directly process streams
+const stream = agent.streamMessage({
+  ...
+});
+
+for await (const update of stream) {
+  ...
+}
+```
+
+**Important:** When using custom implementations, ensure you handle:
+
+- Server-Sent Events (SSE) for `message/stream` and `tasks/resubscribe`
+- Agent card endpoints at `/.well-known/agent-card.json`
+- Proper error handling and JSON-RPC compliance
+- CORS headers if needed for web clients
+
+### Quick-Agents (Alpha v0.5.2)
+
+We are excited to introduce new capabilities for deploying agents directly onto the artinet.
+
+We've added a `testDeployment` utility which is available for all users letting you bundle and test your agent logic in a temporary sandboxed environment.
 
 **QUICK-AGENTS** Use the `fullDeployment` utility, which allows direct deployment of your bundled agent code and `AgentCard` to the Artinet platform (requires an `ARTINET_API_KEY`).
 
@@ -459,89 +698,108 @@ To join the beta waitlist, please email us at humans@artinet.io and stay tuned f
 
 Key features include:
 
--   **Easy Agent Bundling:** Bundle your agent's code and dependencies into a single file using the `bundle` utility.
-    ```typescript
-    import { bundle } from "@artinet/sdk";
+- **Easy Agent Bundling:** Bundle your agent's code and dependencies into a single file using the `bundle` utility from the new deployment export.
 
-    const bundledCode = await bundle(new URL('./your-agent.ts', import.meta.url));
-    ```
+  ```typescript
+  import { bundle } from "@artinet/sdk/deployment";
 
--   **Sandboxed Enviroments:** Streamline agent logic for quick and easy deployments. The new `artinet.v0` namespace (accessible via `@artinet/sdk/agents`) provides `taskManager`, `connect`, and `agent`.
-    -   `artinet.v0.taskManager`: Manages the agent's lifecycle by iterating over the agent's `TaskHandler` and communicating updates to the host environment.
-    -   `artinet.v0.connect`: Replaces the deprecated `fetchResponseProxy`. Allows agents to make proxied calls to other agents or LLMs via the host environment.
-    -   `artinet.v0.agent`: A factory function to obtain a `ClientProxy` for type-safe communication with other agents, managed by the host environment.
+  const bundledCode = await bundle(new URL("./your-agent.ts", import.meta.url));
+  ```
 
-    Example of using the new `artinet.v0` utilities in an agent:
-    ```typescript
-    import { TaskContext, TaskYieldUpdate, Task } from "@artinet/sdk";
-    import { artinet } from "@artinet/sdk/agents";
+- **Sandboxed Enviroments:** Streamline agent logic for quick and easy deployments. The new `artinet.v0` namespace (accessible via `@artinet/sdk/agents`) provides `taskManager`, `connect`, and `agent`.
 
-    export async function* myAgentLogic(context: TaskContext): AsyncGenerator<TaskYieldUpdate, Task | void, unknown> {
-      yield { state: "working" };
+  - `artinet.v0.taskManager`: Manages the agent's lifecycle by iterating over the agent's `TaskHandler` and communicating updates to the host environment.
+  - `artinet.v0.connect`: Replaces the deprecated `fetchResponseProxy`. Allows agents to make proxied calls to other agents or LLMs via the host environment.
+  - `artinet.v0.agent`: A factory function to obtain a `ClientProxy` for type-safe communication with other agents, managed by the host environment.
 
-      // Call another agent/LLM using artinet.v0.connect
-      const llmResponse = await artinet.v0.connect({
-        agentId: "SomeLLMAgentID",
-        messages: [{ role: "user", content: "Tell me a joke." }]
-      });
+  Example of using the new `artinet.v0` utilities in an agent:
 
-      // Or communicate tasks with artinet.v0.agent
-      const anotherAgent = artinet.v0.agent({
-        baseUrl: "https://agents.artinet.io/agentId=456",
-      });
-      const taskResult = await anotherAgent.sendTask({
-        ...
-      });
+  ```typescript
+  import { TaskContext, TaskYieldUpdate, Task } from "@artinet/sdk";
+  import { artinet } from "@artinet/sdk/agents";
 
-    }
+  export async function* myAgentLogic(context: TaskContext): AsyncGenerator<TaskYieldUpdate, Task | void, unknown> {
+    yield { state: "working" };
 
-    // The host environment will invoke this taskManager with the agent's logic.
-    await artinet.v0.taskManager({ taskHandler: myAgentLogic });
-    ```
-    *Note: The `taskHandlerProxy` and `fetchResponseProxy` utilities are now deprecated in favor of `artinet.v0.taskManager` and `artinet.v0.connect` respectively.*
+    // Call another agent/LLM using artinet.v0.connect
+    const llmResponse = await artinet.v0.connect({
+      agentId: "SomeLLMAgentID",
+      messages: [{ role: "user", content: "Tell me a joke." }]
+    });
 
--   **Test-Agents (Experimental):** Simulate and test your agents @ agents.artinet.io/test/deploy using the `testDeployment` tool.
-    ```typescript
-    import { testDeployment, ServerDeploymentRequestParams, SendTaskRequest } from "@artinet/sdk";
+    // Or communicate tasks with artinet.v0.agent
+    const anotherAgent = artinet.v0.agent({
+      baseUrl: "https://agents.artinet.io/agentId=456",
+    });
+    const taskResult = await anotherAgent.sendTask({
+      ...
+    });
 
-    const deploymentParams: ServerDeploymentRequestParams = {
-      code: "/* bundled code string */",
-    };
-    //create a list of tasks for your agent to complete once deployed
-    const testRequests: SendTaskRequest[] = [
-      { id: "t1", message: { role: "user", parts: [{ type: "text", text: "Hi!" }] } }
-    ];
+  }
 
-    for await (const result of testDeployment(deploymentParams, testRequests)) {
-      console.log(result); //process the task completion requests as they come in to confirm your agents logic
-    }
-    ```
+  // The host environment will invoke this taskManager with the agent's logic.
+  await artinet.v0.taskManager({ taskHandler: myAgentLogic });
+  ```
 
--   **Full Deployment (Experimental):** Deploy your agent to the Artinet platform using the `fullDeployment` utility.
-    ```typescript
-    import { fullDeployment, ServerDeploymentRequestParams } from "@artinet/sdk";
+  _Note: The `taskHandlerProxy` and `fetchResponseProxy` utilities are now deprecated in favor of `artinet.v0.taskManager` and `artinet.v0.connect` respectively._
 
-    const deploymentParams: ServerDeploymentRequestParams = {
-      name: "My Awesome Agent",
-      agentCard: { /* your agent card */ },
-      code: "/* bundled code string */",
-    };
+- **Test-Agents (Experimental):** Simulate and test your agents @ agents.artinet.io/test/deploy using the `testDeployment` tool.
+
+  ```typescript
+  import {
+    testDeployment,
+    ServerDeploymentRequestParams,
+    SendTaskRequest,
+  } from "@artinet/sdk/deployment";
+
+  const deploymentParams: ServerDeploymentRequestParams = {
+    code: "/* bundled code string */",
+  };
+  //create a list of tasks for your agent to complete once deployed
+  const testRequests: SendTaskRequest[] = [
+    {
+      id: "t1",
+      message: { role: "user", parts: [{ type: "text", text: "Hi!" }] },
+    },
+  ];
+
+  for await (const result of testDeployment(deploymentParams, testRequests)) {
+    console.log(result); //process the task completion requests as they come in to confirm your agents logic
+  }
+  ```
+
+- **Full Deployment (Experimental):** Deploy your agent to the Artinet platform using the `fullDeployment` utility.
+
+  ```typescript
+  import {
+    fullDeployment,
+    ServerDeploymentRequestParams,
+  } from "@artinet/sdk/deployment";
+
+  const deploymentParams: ServerDeploymentRequestParams = {
+    name: "My Awesome Agent",
+    agentCard: {
+      /* your agent card */
+    },
+    code: "/* bundled code string */",
+  };
+
+  const deploymentResult = await fullDeployment(deploymentParams); // Requires an ARTINET_API_KEY environment variable
+  console.log("Deployment Result:", deploymentResult);
+  ```
+
+- **Dedicated Endpoints:** Once deployed your agent will be available On-Demand at its dedicated enpoint. (e.g. "https://agents.artinet.io/agentId=0xabf698845743538727a81352bfcfdb724e5c2bbe3113a26362482248f9f3e5fa/.well-known/agent-card.json")
+- **New Types:** To support these features, new types for server deployment requests and responses (such as `ServerDeploymentRequestParams`, `ServerDeploymentResponse`) have been added to `src/types/extended-schema.ts`. New types for sandboxed agent interactions (`TaskProxy`, `ConnectAPICallback`, `ClientProxy`, etc.) are in `src/types/proxy.ts`.
+
+**QUICK-AGENT FAQs**
+
+- Test-Agents expire after 60s (need more time? let us know @humans@artinet.io)
+- Quick-Agents do not have access to a filesystem or networking (limited persistance & networking capabalities are on the project roadmap).
+- Quick-Agents v0 does not support streaming, push notifications or state transition history (these capabilities are on the project roadmap).
+- Larger deployments can take significant time to deploy which may cause `fullDeployment` to timeout. In such cases wait to see if the listing has been added to your account before trying to deploy again.
+- Quick-Agent logic is public, therefore the artinet project is not liable for any sensitive material held within a deployment.
+- Available with version 0.5.6+ of the SDK with enhanced deployment capabilities.
 
-    const deploymentResult = await fullDeployment(deploymentParams); // Requires an ARTINET_API_KEY environment variable
-    console.log("Deployment Result:", deploymentResult);
-    ```
-
--   **Dedicated Endpoints:** Once deployed your agent will be available On-Demand at its dedicated enpoint. (e.g. "https://agents.artinet.io/agentId=0xabf698845743538727a81352bfcfdb724e5c2bbe3113a26362482248f9f3e5fa/.well-known/agent.json")
--   **New Types:** To support these features, new types for server deployment requests and responses (such as `ServerDeploymentRequestParams`, `ServerDeploymentResponse`) have been added to `src/types/extended-schema.ts`. New types for sandboxed agent interactions (`TaskProxy`, `ConnectAPICallback`, `ClientProxy`, etc.) are in `src/types/proxy.ts`.
-
-  **QUICK-AGENT FAQs**
--   Test-Agents expire after 60s (need more time? let us know @humans@artinet.io)
--   Quick-Agents do not have access to a filesystem or networking (limited persistance & networking capabalities are on the project roadmap).
--   Quick-Agents v0 does not support streaming, push notifications or state transition history (these capabilities are on the project roadmap).
--   Larger deployments can take significant time to deploy which may cause `fullDeployment` to timeout. In such cases wait to see if the listing has been added to your account before trying to deploy again.
--   Quick-Agent logic is public, therefore the artinet project is not liable for any sensitive material held within a deployment.
--   Only availble with version 0.5.3 of the SDK.
-  
 Sign-up at [artinet.io](https://artinet.io/) to deploy your Quick-Agent today!
 
 ## Contributing
@@ -561,6 +819,5 @@ This SDK builds upon the concepts and specifications of the [Agent2Agent (A2A) P
 Libraries used include:
 
 - [Express.js](https://expressjs.com/) for the server framework.
-- [Jayson](https://github.com/tedeh/jayson) for JSON-RPC handling.
 - [Pino](https://getpino.io/) for logging.
 - [EventSource Parser](https://github.com/rexxars/eventsource-parser) for SSE streaming.