AgentCore Developer Guide — Complete reference for Runtime, Memory, Gateway, Identity, Observability, Evaluations, Policy, and Harness
AgentCore CLI — CLI source, release notes, and the authoritative agentcore.json schema reference
AgentCore Samples — End-to-end reference implementations covering workflow, conversational, and multi-agent patterns
AgentCore Code Interpreter — Managed, isolated code execution environment with Sandbox, Public, and VPC network modes
AgentCore Browser Tool — Managed browser sandbox for agents that interact with web applications, with AWS managed and custom browser options
AgentCore CloudFormation Resources — CloudFormation resource types for provisioning AgentCore infrastructure as code
AgentCore CDK Constructs — AWS CDK L2 construct library for AgentCore resources
AgentCore Terraform Provider — Terraform provider resources for AgentCore agent management

Existing-Agent Adoption

Runtime Lifecycle Settings — Per-session microVM lifecycle, idle timeouts, and maximum lifetimes
AgentCore Pricing — Active CPU and memory consumption rates and associated Runtime costs
Configure AgentCore Observability — Framework instrumentation, ADOT, trace destinations, and CloudWatch configuration
Configure Runtime for VPC — Private-resource connectivity, ENIs, security groups, endpoints, and routing
Runtime Security Best Practices — Session isolation, least-privilege roles, network controls, and credential guidance
Runtime Credential Management (MMDS) — How code in Runtime microVMs obtains temporary execution-role credentials
Cost-Efficient Initialization and Scale-to-Zero — Well-Architected guidance for lifecycle settings, session reuse, and initialization caching

Architecture and Best Practices

Well-Architected Agentic AI Lens — Security, reliability, and operational best practices for agentic AI workloads
Well-Architected Generative AI Lens — Cost optimization, performance, and governance guidance for generative AI
Agentic AI Security Scoping Matrix — Risk classification framework mapping agent autonomy levels to security controls
Security Best Practices for Agentic AI — AWS Prescriptive Guidance for secure agentic AI system design, development, governance, and incident response
Building Serverless Architectures for Agentic AI — Prescriptive guidance for modular, event-driven AI architectures using serverless AWS services
Memory-Augmented Agents — Prescriptive guidance for agents that retain and reason over short- and long-term context
Cost-Efficient Agent Initialization — Well-Architected guidance on Runtime session lifecycle, scale-to-zero behavior, and initialization caching
Circuit Breaker Pattern — Fault-tolerance pattern that prevents cascading failures when a downstream service is unhealthy
Cedar Policy Language — Authorization policy language used by AgentCore for tool and resource access control

Strands Agents Python Quickstart — Official SDK guide for building a first agent with models and tools
Operating Agents in Production — Monitoring, scaling, and reliability guidance for deployed agents
Deploying Strands Agents to AgentCore — Deploy Strands agents on AgentCore Runtime with a managed, serverless runtime
Swarm Multi-Agent Pattern — Collaborative multi-agent orchestration with autonomous coordination and shared context
Introducing Strands Agents — Overview of Strands’ model-driven approach to building agents with tools

OWASP Top 10 for LLM Applications — Industry-standard vulnerability taxonomy for LLM-powered systems
NIST AI 600-1: Generative AI Profile — NIST guidance for managing generative AI risks alongside the AI Risk Management Framework
Cedar Policy Language Reference — Specification and examples for writing Cedar authorization policies
MicroVM Metadata Service (MMDS) — How AgentCore Runtime exposes execution-role credentials inside microVMs and the least-privilege implications
Context Window Overflow in Agents — Security considerations for mitigating context-window overflow in agentic systems
Prompt Injection Mitigation — Defense patterns for protecting agents from indirect prompt injection
Safeguarding GenAI Workloads from Prompt Injections — AWS Security Blog guidance on direct and indirect prompt injection defense
OWASP GenAI Security Project — Community hub for generative AI security guidance, taxonomies, and testing resources
SaaS Tenant Isolation Strategies — AWS whitepaper on isolation models from pooled infrastructure to full-stack separation

Amazon EventBridge for Event-Driven Agents — Serverless event bus for decoupling agent triggers from processing logic
Amazon SQS Dead-Letter Queues — Capture and retry failed agent invocations without data loss
Step Functions Error Handling — Retry, catch, and fallback patterns for multi-step agent workflows
Human Approval with Step Functions — Callback patterns for pausing workflows pending human review
Lambda Durable Functions — Checkpoint-and-replay durable execution (up to a year) for long-running, fault-tolerant multi-step agent orchestration expressed as code
Async AgentCore with Step Functions and durable functions — Sample comparing Step Functions callbacks and a durable function for invoking an AgentCore agent without blocking a Lambda during processing
Tenant Isolation with IAM — Resource and policy isolation patterns for multi-tenant agent deployments
Saga Pattern for Distributed Transactions — Compensating transactions for reliable multi-service agent operations

BGL: Democratizing Business Intelligence — How BGL built a Claude Agent SDK and AgentCore solution for self-service analytics
Thomson Reuters: Legal AI — Document analysis agents for legal research at enterprise scale
Iberdrola: Energy Operations — Predictive maintenance agents managing grid infrastructure
AutoScout24: Bot Factory — Standardized, secure agent development through reusable templates and event-driven services
Caylent: Cloud Cost Analysis — CloudZero Advisor uses specialized agents to provide cloud cost analysis through natural-language interaction