adding typescript (#1408)

Author: evandrofranco

01-tutorials/01-AgentCore-runtime/01-hosting-agent/07-direct-code-deploy-typescript/.gitignore

@@ -0,0 +1,20 @@
+__pycache__
+*.pyc
+.venv/
+.env
+.ipynb_checkpoints/
+.kiro/
+.vscode/
+.DS_Store
+tmp/
+**/chat_app/
+**/travel_chat/
+dist/
+node_modules
+my-agent.zip
+
+
+# to-remove
+models/
+config.py
+travel_chat

01-tutorials/01-AgentCore-runtime/01-hosting-agent/07-direct-code-deploy-typescript/README.md

@@ -0,0 +1,224 @@
+# Getting Started — TypeScript Agent on Amazon Bedrock AgentCore
+
+Deploy a TypeScript agent to AgentCore Runtime using Direct Code Deploy with Node.js 22.
+
+## Prerequisites
+
+| Requirement | Version | Install |
+|---|---|---|
+| Node.js | 22.x | [nodejs.org](https://nodejs.org/) |
+| AWS CLI | 2.x | [AWS CLI install guide](https://docs.aws.amazon.com/cli/latest/userguide/getting-started-install.html) |
+| jq | latest | `brew install jq` / `apt install jq` |
+
+### Configure AWS credentials
+
+```bash
+aws configure
+# Or set environment variables:
+# export AWS_ACCESS_KEY_ID=<your-key>
+# export AWS_SECRET_ACCESS_KEY=<your-secret>
+# export AWS_DEFAULT_REGION=us-west-2
+```
+
+Verify your credentials:
+
+```bash
+aws sts get-caller-identity
+```
+
+---
+
+## Project Structure
+
+```
+typescript/
+├── app.ts               # Agent entry point (Express server)
+├── package.json          # Node.js dependencies
+├── tsconfig.json         # TypeScript config
+├── iam.sh                # IAM role creation/deletion (bash + AWS CLI)
+├── runtime.sh            # Create, get, list, wait, invoke, and delete runtimes (bash + AWS CLI)
+└── README.md
+```
+
+---
+
+## Step 1: Create the IAM Execution Role
+
+AgentCore Runtime needs an IAM role to run your agent. The `iam.sh` script creates a role named `TypescriptExecutionRole` with the necessary permissions (Bedrock model invocation, ECR pull, CloudWatch Logs, X-Ray, and AgentCore services).
+
+```bash
+./iam.sh create
+```
+
+This is idempotent — if the role already exists, it returns the existing ARN. Capture the output and export it:
+
+```bash
+export ROLE_ARN=$(./iam.sh create)
+echo $ROLE_ARN
+```
+
+---
+
+## Step 2: Build the Agent Package
+
+Install dependencies, bundle TypeScript + all dependencies into a single file, and create the deployment zip:
+
+```bash
+npm install
+
+npm run build
+
+cd dist
+zip deployment_package.zip app.js
+cd ..
+
+```
+
+This uses `esbuild` to bundle `app.ts` + all dependencies (Express, Strands Agents SDK, Zod, AWS SDK) into a single `dist/app.js`. No `node_modules` needed in the zip.
+
+---
+
+## Step 3: Upload to S3
+
+Set your account ID and region, then upload:
+
+```bash
+export BUCKET=$(your-bucket)
+
+aws s3 cp dist/deployment_package.zip \
+  s3://$BUCKET/typescript_deploy/deployment_package.zip
+```
+
+---
+
+## Step 4: Deploy with Direct Code Deploy
+
+Make sure `ROLE_ARN` is exported from Step 1, then create the runtime:
+
+```bash
+export AWS_REGION="us-east-1"
+# ROLE_ARN already exported from Step 1
+
+./runtime.sh create
+```
+
+Example output:
+
+```json
+{
+  "agentRuntimeArn": "arn:aws:bedrock-agentcore:us-west-2:123456789012:runtime/my_typescript_agent-XXXXXXX",
+  "agentRuntimeId": "my_typescript_agent-XXXXXXX",
+  "status": "CREATING"
+}
+```
+
+Export your agent ARN to be referenced in next examples:
+
+```bash
+export AGENT_ARN="arn:aws:bedrock-agentcore:us-west-2:123456789012:runtime/my_typescript_agent-XXXXXXX"
+export AGENT_ID="my_typescript_agent-XXXXXXX"
+```
+
+
+Wait for the runtime to become `READY` (polls every 10s):
+
+```bash
+./runtime.sh wait $AGENT_ID
+```
+
+---
+
+## Step 5: Verify and Invoke
+
+### List runtimes
+
+```bash
+./runtime.sh list
+```
+
+### Get runtime details
+
+```bash
+./runtime.sh get $AGENT_ID
+```
+
+### Invoke the agent
+
+Once the runtime status is `READY`:
+
+```bash
+./runtime.sh invoke $AGENT_ARN
+
+# With a custom prompt:
+./runtime.sh invoke $AGENT_ARN "what is your status?"
+```
+
+---
+
+## Step 6: Clean Up
+
+Delete the runtime, IAM role, and S3 artifact when you're done:
+
+```bash
+# Delete the runtime
+./runtime.sh delete <agentRuntimeId>
+
+# Delete the IAM role
+./iam.sh delete
+
+# Remove the S3 artifact
+aws s3 rm s3://bedrock-agentcore-code-${ACCOUNT_ID}-${REGION}/typescript_deploy/deployment_package.zip
+```
+
+---
+
+## How It Works
+
+### Agent Code (`app.ts`)
+
+The agent is an Express server powered by [Strands Agents SDK](https://strandsagents.com/) with a **calculator tool**. The Strands agent uses Amazon Bedrock (Claude Haiku 4.5) as the LLM and can call tools via native tool-calling.
+
+| Endpoint | Method | Purpose |
+|---|---|---|
+| `/ping` | GET | Health check — AgentCore uses this to verify the agent is running |
+| `/invocations` | POST | Receives a prompt, runs the Strands agent (with tool use), and returns the response |
+
+AgentCore Runtime expects the server to listen on port **8080**.
+
+#### Calculator Tool
+
+The agent has a `calculator` tool that supports `add`, `subtract`, `multiply`, and `divide`. When you send a math question, the LLM decides to call the tool and returns the result.
+
+```bash
+./runtime.sh invoke $AGENT_ARN "What is 25 * 4 + 10?"
+```
+
+### Direct Code Deploy
+
+Instead of building a container image, Direct Code Deploy lets you upload your source code as a zip to S3. AgentCore handles the build and runtime environment. The deploy payload specifies:
+
+```json
+{
+  "agentRuntimeArtifact": {
+    "codeConfiguration": {
+      "code": {
+        "s3": {
+          "bucket": "bedrock-agentcore-code-<ACCOUNT_ID>-<REGION>",
+          "prefix": "typescript_deploy/deployment_package.zip"
+        }
+      },
+      "runtime": "NODE_22",
+      "entryPoint": ["app.js"]
+    }
+  }
+}
+```
+
+### IAM Role
+
+The execution role grants the agent permissions to:
+- Invoke Bedrock models (`bedrock:InvokeModel`)
+- Pull container images from ECR
+- Write logs to CloudWatch
+- Send traces to X-Ray
+- Access AgentCore services (Memory, Browser, Gateway, CodeInterpreter)

01-tutorials/01-AgentCore-runtime/01-hosting-agent/07-direct-code-deploy-typescript/app.ts

@@ -0,0 +1,69 @@
+import http from "node:http";
+import zlib from "node:zlib";
+import { Agent, tool } from "@strands-agents/sdk";
+import { BedrockModel } from "@strands-agents/sdk/models/bedrock";
+import z from "zod";
+
+const model = new BedrockModel({
+  region: process.env.AWS_REGION || "us-east-1",
+  modelId: "global.anthropic.claude-haiku-4-5-20251001-v1:0",
+});
+
+const calculator = tool({
+  name: "calculator",
+  description: "Perform basic arithmetic: add, subtract, multiply, divide.",
+  inputSchema: z.object({
+    operation: z.enum(["add", "subtract", "multiply", "divide"]),
+    a: z.number(),
+    b: z.number(),
+  }),
+  callback: ({ operation, a, b }) => {
+    if (operation === "add") return `${a + b}`;
+    if (operation === "subtract") return `${a - b}`;
+    if (operation === "multiply") return `${a * b}`;
+    if (operation === "divide") return b === 0 ? "Error: division by zero" : `${a / b}`;
+    return "Unknown operation";
+  },
+});
+
+const agent = new Agent({ model, tools: [calculator] });
+
+function readBody(req: http.IncomingMessage): Promise<string> {
+  return new Promise((resolve, reject) => {
+    const chunks: Buffer[] = [];
+    const stream = req.headers["content-encoding"] === "gzip" ? req.pipe(zlib.createGunzip()) : req;
+    stream.on("data", (c: Buffer) => chunks.push(c));
+    stream.on("end", () => resolve(Buffer.concat(chunks).toString("utf-8")));
+    stream.on("error", reject);
+  });
+}
+
+http.createServer(async (req, res) => {
+  if (req.url === "/ping") {
+    res.writeHead(200, { "Content-Type": "application/json" });
+    return res.end('{"status":"Healthy"}');
+  }
+
+  if (req.method === "POST" && req.url === "/invocations") {
+    let prompt = "Hello";
+    try {
+      const raw = await readBody(req);
+      try {
+        const body = JSON.parse(raw);
+        if (body.prompt) prompt = body.prompt;
+      } catch {
+        const text = raw.trim();
+        if (text) prompt = text;
+      }
+    } catch {
+      // use default prompt
+    }
+
+    const result = await agent.invoke(prompt);
+    res.writeHead(200, { "Content-Type": "application/json" });
+    return res.end(JSON.stringify({ response: result.lastMessage }));
+  }
+
+  res.writeHead(404);
+  res.end();
+}).listen(8080, "0.0.0.0");