Bedrock AgentCore SDK

Deploy your local AI agent to Bedrock AgentCore with zero infrastructure

Python SDK ◆ Starter Toolkit ◆ Samples ◆ Discord

🚀 From Local Development to Bedrock AgentCore

# Your existing agent (any framework)
from strands import Agent
# or LangGraph, CrewAI, Autogen, custom logic - doesn't matter

def my_local_agent(query):
    # Your carefully crafted agent logic
    return agent.process(query)

# Deploy to Bedrock AgentCore
from bedrock_agentcore import BedrockAgentCoreApp
app = BedrockAgentCoreApp()

@app.entrypoint
def production_agent(request):
    return my_local_agent(request.get("prompt"))  # Same logic, enterprise platform

app.run()  # Ready to run on Bedrock AgentCore

What you get with Bedrock AgentCore:

• ✅ Keep your agent logic - Works with Strands, LangGraph, CrewAI, Autogen, custom frameworks
• ✅ Zero infrastructure management - No servers, containers, or scaling concerns
• ✅ Enterprise-grade platform - Built-in auth, memory, observability, security
• ✅ Production-ready deployment - Reliable, scalable, compliant hosting

⚠️ Preview Status

Bedrock AgentCore SDK is currently in public preview. APIs may change as we refine the SDK.

🛠️ Built for AI Developers

Real-time Health Monitoring

@app.async_task  # Automatically tracks background work
async def process_documents(files):
    # Long-running AI processing
    return results

@app.ping  # Custom health status
def health_check():
    return "HEALTHY" if all_services_up() else "HEALTHY_BUSY"

Enterprise Platform Services

• 🧠 Memory - Persistent knowledge across sessions
• 🔗 Gateway - Transform APIs into MCP tools
• 💻 Code Interpreter - Secure sandboxed execution
• 🌐 Browser - Cloud-based web automation
• 📊 Observability - OpenTelemetry tracing
• 🔐 Identity - AWS & third-party auth

🏗️ Deployment

Quick Start: Use the Bedrock AgentCore Starter Toolkit for rapid prototyping.

Production: AWS CDK - coming soon.

📝 License & Contributing

• License: Apache 2.0 - see LICENSE.txt
• Contributing: See CONTRIBUTING.md
• Security: Report vulnerabilities via SECURITY.md