feat: add aws bedrock agentcore examples (#811)

Author: czubocha

aws-bedrock-agentcore/javascript/langgraph-basic-dockerfile/Dockerfile

@@ -0,0 +1,48 @@
+# LangGraph JS Minimal Agent - Dockerfile
+#
+# Multi-stage build for smaller production image.
+# Uses Node.js 20 LTS with ES Modules support.
+
+# =============================================================================
+# Stage 1: Dependencies
+# =============================================================================
+FROM node:20-slim AS deps
+
+WORKDIR /app
+
+# Copy package files
+COPY package*.json ./
+
+# Install production dependencies only
+RUN npm ci --omit=dev
+
+# =============================================================================
+# Stage 2: Production
+# =============================================================================
+FROM node:20-slim AS production
+
+WORKDIR /app
+
+# Create non-root user for security
+RUN groupadd --gid 1001 nodejs && \
+    useradd --uid 1001 --gid nodejs --shell /bin/bash --create-home agent
+
+# Copy dependencies from deps stage
+COPY --from=deps /app/node_modules ./node_modules
+
+# Copy application code
+COPY agent.js ./
+COPY package.json ./
+
+# Set environment variables
+ENV NODE_ENV=production
+ENV PORT=8080
+
+# Switch to non-root user
+USER agent
+
+# Expose the AgentCore runtime port
+EXPOSE 8080
+
+# Start the agent
+CMD ["node", "agent.js"]

aws-bedrock-agentcore/javascript/langgraph-basic-dockerfile/README.md

@@ -0,0 +1,187 @@
+# LangGraph JS Minimal Example
+
+A minimal LangGraph JavaScript agent deployed to AWS Bedrock AgentCore using Serverless Framework.
+
+## Features
+
+- **LangGraph JS**: ReAct agent pattern with tool calling
+- **Claude Sonnet 4.5**: Powered by Amazon Bedrock
+- **Simple Tools**: Calculator operations and time queries
+- **Docker Deployment**: Auto-detected Dockerfile for easy deployment
+
+## Quick Start
+
+### Prerequisites
+
+- Node.js 20+
+- Docker (for local development)
+- AWS credentials configured
+- Serverless Framework CLI (`npm install -g serverless`)
+
+### Install Dependencies
+
+```bash
+npm install
+```
+
+### Local Development
+
+Start the agent locally with hot reload:
+
+```bash
+sls dev
+```
+
+This will:
+
+1. Build the Docker image
+2. Start the container with AWS credentials injected
+3. Open an interactive chat interface
+4. Watch for file changes and auto-rebuild
+
+Try these prompts:
+
+- "What time is it?"
+- "What time is it in Tokyo?"
+- "Calculate 25 multiplied by 4"
+- "Add 100 and 250, then divide the result by 7"
+
+### Deploy to AWS
+
+```bash
+sls deploy
+```
+
+### Invoke Deployed Agent
+
+```bash
+# Via Serverless Framework CLI
+sls invoke --agent assistant --data '{"prompt":"Hello! What can you help me with?"}'
+
+# Plain string is also supported:
+# sls invoke --agent assistant -d "Hello! What can you help me with?"
+
+# Or via curl (replace URL with your runtime URL from `sls info`)
+curl -X POST https://your-runtime-url/invoke \
+  -H "Content-Type: application/json" \
+  -d '{"prompt":"Hello! What can you help me with?"}'
+```
+
+### Remove
+
+```bash
+sls remove
+```
+
+## Project Structure
+
+```text
+langgraph-basic-dockerfile/
+├── serverless.yml    # Serverless Framework configuration
+├── agent.js          # LangGraph JS agent with tools
+├── package.json      # npm dependencies
+├── Dockerfile        # Container definition
+└── README.md         # This file
+```
+
+## How It Works
+
+### Agent Architecture
+
+```text
+User Input
+    │
+    ▼
+┌─────────────────┐
+│  BedrockAgent   │
+│    CoreApp      │◄─── HTTP Server (port 8080)
+└────────┬────────┘
+         │
+         ▼
+┌─────────────────┐
+│   LangGraph     │
+│  ReAct Agent    │◄─── Alternates between LLM and tools
+└────────┬────────┘
+         │
+    ┌────┴────┐
+    ▼         ▼
+┌───────┐ ┌───────┐
+│ Tools │ │ Claude│
+│       │ │  LLM  │
+└───────┘ └───────┘
+```
+
+### Tools Available
+
+| Tool               | Description                                  |
+| ------------------ | -------------------------------------------- |
+| `get_current_time` | Get current date/time with optional timezone |
+| `add`              | Add two numbers                              |
+| `multiply`         | Multiply two numbers                         |
+| `divide`           | Divide two numbers                           |
+
+### Configuration
+
+The `serverless.yml` is intentionally minimal:
+
+```yaml
+service: langgraph-basic-dockerfile
+
+provider:
+  name: aws
+
+ai:
+  agents:
+    assistant: {}
+```
+
+The Dockerfile is auto-detected, and default settings are applied:
+
+- Protocol: HTTP
+- Network: PUBLIC
+- Port: 8080
+
+## Customization
+
+### Adding New Tools
+
+Edit `agent.js` to add new tools:
+
+```javascript
+const myNewTool = tool(
+  async ({ input }) => {
+    // Tool implementation
+    return `Result: ${input}`
+  },
+  {
+    name: 'my_new_tool',
+    description: 'Description for the LLM',
+    schema: z.object({
+      input: z.string().describe('Input parameter'),
+    }),
+  },
+)
+
+// Add to tools array
+const tools = [getCurrentTime, add, multiply, divide, myNewTool]
+```
+
+### Changing the Model
+
+Edit the model configuration in `agent.js`:
+
+```javascript
+const model = new ChatBedrockConverse({
+  // Claude Sonnet 4.5
+  model: 'us.anthropic.claude-sonnet-4-5-20250929-v1:0',
+  // Or use other models:
+  // model: 'us.amazon.nova-2-lite-v1:0',
+  // model: 'us.meta.llama3-70b-instruct-v1:0',
+  region: process.env.AWS_REGION || 'us-east-1',
+})
+```
+
+## Related Examples
+
+- [langgraph-basic-docker](../../python/langgraph-basic-docker/) - Python version
+- [langgraph-gateway](../../python/langgraph-gateway/) - Lambda functions as tools

aws-bedrock-agentcore/javascript/langgraph-basic-dockerfile/agent.js

@@ -0,0 +1,152 @@
+/**
+ * Minimal LangGraph JS agent with simple built-in tools.
+ *
+ * This agent demonstrates:
+ * - BedrockAgentCoreApp entrypoint pattern for JavaScript
+ * - LangChain createAgent (backed by LangGraph) with Claude Sonnet 4.5 via Bedrock
+ * - Simple tool integration (calculator, time)
+ * - Docker-based deployment
+ */
+
+import { BedrockAgentCoreApp } from 'bedrock-agentcore/runtime'
+import { createAgent } from 'langchain'
+import { ChatBedrockConverse } from '@langchain/aws'
+import { tool } from '@langchain/core/tools'
+import { z } from 'zod'
+
+// Initialize Claude Sonnet 4.5 via US inference profile
+const model = new ChatBedrockConverse({
+  model: 'us.anthropic.claude-sonnet-4-5-20250929-v1:0',
+  region: process.env.AWS_REGION || 'us-east-1',
+})
+
+// Define simple tools using the tool() helper
+
+/**
+ * Get the current date and time
+ */
+const getCurrentTime = tool(
+  async ({ timezone }) => {
+    const now = new Date()
+    const options = {
+      timeZone: timezone || 'UTC',
+      dateStyle: 'full',
+      timeStyle: 'long',
+    }
+    return `Current time: ${now.toLocaleString('en-US', options)}`
+  },
+  {
+    name: 'get_current_time',
+    description:
+      'Get the current date and time. Optionally specify a timezone.',
+    schema: z.object({
+      timezone: z
+        .string()
+        .optional()
+        .describe(
+          'Timezone (e.g., "America/New_York", "Europe/London", "UTC")',
+        ),
+    }),
+  },
+)
+
+/**
+ * Add two numbers together
+ */
+const add = tool(
+  async ({ a, b }) => {
+    const result = a + b
+    return `${a} + ${b} = ${result}`
+  },
+  {
+    name: 'add',
+    description: 'Add two numbers together.',
+    schema: z.object({
+      a: z.number().describe('First number'),
+      b: z.number().describe('Second number'),
+    }),
+  },
+)
+
+/**
+ * Multiply two numbers together
+ */
+const multiply = tool(
+  async ({ a, b }) => {
+    const result = a * b
+    return `${a} × ${b} = ${result}`
+  },
+  {
+    name: 'multiply',
+    description: 'Multiply two numbers together.',
+    schema: z.object({
+      a: z.number().describe('First number'),
+      b: z.number().describe('Second number'),
+    }),
+  },
+)
+
+/**
+ * Divide two numbers
+ */
+const divide = tool(
+  async ({ a, b }) => {
+    if (b === 0) {
+      return 'Error: Cannot divide by zero'
+    }
+    const result = a / b
+    return `${a} ÷ ${b} = ${result}`
+  },
+  {
+    name: 'divide',
+    description: 'Divide two numbers.',
+    schema: z.object({
+      a: z.number().describe('Dividend (number to divide)'),
+      b: z.number().describe('Divisor (number to divide by)'),
+    }),
+  },
+)
+
+// Collect all tools
+const tools = [getCurrentTime, add, multiply, divide]
+
+// Create LangGraph agent
+// This creates a graph that alternates between calling the LLM and executing tools
+const agent = createAgent({
+  model,
+  tools,
+})
+
+// Initialize the AgentCore application with invocation handler
+const app = new BedrockAgentCoreApp({
+  invocationHandler: {
+    // Define the expected request schema
+    requestSchema: z.object({
+      prompt: z.string().describe('The user message to process'),
+    }),
+
+    // Process incoming requests (non-streaming)
+    async process(request, context) {
+      console.log(`Received message: ${request.prompt}`)
+      console.log(`Request ID: ${context?.requestId || 'unknown'}`)
+
+      // Invoke the LangGraph agent with the user message
+      const result = await agent.invoke({
+        messages: [{ role: 'user', content: request.prompt }],
+      })
+
+      // Extract the final message from the graph result
+      const finalMessage = result.messages[result.messages.length - 1]
+      const response = finalMessage.content
+
+      console.log(`Responding with: ${response}`)
+
+      // Return the response (non-streaming)
+      return response
+    },
+  },
+})
+
+// Start the AgentCore application
+// This starts an HTTP server on port 8080 (default)
+app.run()