Fullstack Solution Template for AgentCore (FAST)

A flexible fullstack solution template for production-ready deployments on Amazon Bedrock AgentCore, featuring pre-built frontend/backend, secure authentication, and CDK infrastructure, compatible with Strands, LangGraph, and other Agent SDKs.

Overview#

Fullstack Solution Template for AgentCore (FAST) is an officially maintained solution template by AWS Labs, designed to accelerate production-ready Agent application deployments on Amazon Bedrock AgentCore. Current version: 0.3.1, licensed under Apache-2.0.

Key Features#

Fullstack out-of-the-box: React frontend + AgentCore Runtime backend + CDK infrastructure
Secure authentication: Four-layer Amazon Cognito auth (frontend login, Runtime access, Gateway access, API Gateway)
Agent patterns: strands-single-agent (default), langgraph-single-agent
Built-in tools: Gateway Tools (Lambda-powered), Code Interpreter (sandboxed Python execution)
Vibe-Coding friendly: Supports Amazon Q, Kiro, Cline, Claude Code and other AI coding assistants

Architecture#

Project Structure#

fullstack-agentcore-solution-template/
├── frontend/           # React frontend application
├── infra-cdk/          # CDK infrastructure code
├── patterns/           # Agent pattern implementations
├── tools/              # Reusable tools (framework-agnostic)
├── gateway/            # Gateway tools and utilities
├── scripts/            # Deployment and test scripts
├── vibe-context/       # AI coding assistant context
└── docker/             # Local development Docker config

Key Mechanisms#

AgentCore Memory: Stores conversation history for multi-turn context
AgentCore Gateway: Provides tool access via MCP
Code Interpreter: Sandboxed execution with pre-installed Python libraries

Installation & Deployment#

Prerequisites#

Node.js 20+
Python 3.11+
AWS CLI configured with credentials
AWS CDK CLI
Docker (required)

Quick Deploy#

cd infra-cdk
npm install
cdk bootstrap  # First-time deployment
cdk deploy
cd ..
python scripts/deploy-frontend.py

Configuration Example#

stack_name_base: your-project-name  # Max 35 chars
admin_user_email: null              # Optional
backend:
  pattern: strands-single-agent     # Options: strands-single-agent, langgraph
  deployment_type: docker           # Options: docker (default), zip

Use Cases#

Multi-turn conversational Agent applications
Tool-calling Agent systems
Production-grade secure deployment
Rapid AgentCore solution validation

Important Notes#

Project documentation states "proof-of-value" - users must implement their own security controls
Currently only supports Single-Agent mode, Multi-Agent planned for future versions
Non-ARM deployment requires QEMU and Docker buildx configuration