Amazon Bedrock AgentCore reached general availability on October 13, 2025, in nine AWS regions. It is a set of AWS services for building, deploying, and operating AI agents in production: Runtime, Memory, Gateway, Identity, Code Interpreter, Browser, Observability, Evaluations, Policy, Registry, Payments, and A2A, thirteen services in total. Two of them, Registry and Payments, are currently in preview. Since launch, regional coverage has expanded to 20 regions.
AgentCore is framework-agnostic. It works with Strands Agents, LangGraph, CrewAI, LlamaIndex, and Google ADK, and with any underlying model. Pricing is consumption-based, with no upfront commitments or minimum fees. Each agent session runs in a dedicated microVM, and memory is sanitized at session termination. Two protocols connect the pieces: MCP for tools, and A2A, an open agent-to-agent standard under the Linux Foundation, for agent-to-agent communication.
That is a lot of surface area for a platform less than a year old. Most of what exists to learn it today is the official AWS documentation itself, plus scattered blog posts covering one service at a time.
What this is
AWS AgentCore Blueprints is a 36-chapter book covering the platform end to end, published as free visual chapters (engineering-sheet style pages) at tocconsulting.fr/agentcore.
It is organized in five parts:
Part I, Foundations: What AgentCore is, the terminology, the common mistakes teams make early on, and what happens inside an agent (the harness, and how Strands Agents powers it).
Part II, Core Services: Runtime, Memory, Gateway, and Identity: hosting, remembering, tools, and secure access.
Part III, Tools and Operations: Code Interpreter, Browser, Observability, and Evaluations: what an agent can do, and how you watch it do it.
Part IV, Governance and Transactions: Policy, Registry, and Payments: rules, catalogs, and agents that spend real money through the x402 payment protocol.
Part V, Multi-Agent and Architecture: A2A and multi-agent patterns, architecture decisions, multi-tenancy, a threat model built on the OWASP Top 10 for LLM Applications, org-scale governance across accounts, and multi-region design.
Each concept chapter is paired with a hands-on practical chapter: CLI commands, SDK calls, and code, written to match the official AWS quickstarts and the amazon-bedrock-agentcore-samples repository.
How it was built
The book follows the AWS documentation and the official AWS samples repository, chapter by chapter, with links to the AWS pages throughout.
AgentCore changes frequently. Some details in the book, region lists and preview statuses in particular, will drift over time. Anyone building on this should always confirm current limits and status against the AWS documentation directly.
Read it
The book is free, with no email required: https://tocconsulting.fr/agentcore
References
- AgentCore general availability, October 13, 2025, nine regions: https://aws.amazon.com/about-aws/whats-new/2025/10/amazon-bedrock-agentcore-available/
- AgentCore overview and core services: https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/what-is-bedrock-agentcore.html
- Region list (20 regions as of the June 2026 expansion): https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/agentcore-regions.html and https://aws.amazon.com/about-aws/whats-new/2026/06/amazon-bedrock-agentcore-four-additional-regions/
- A2A as an open, Linux Foundation standard: https://aws.amazon.com/blogs/opensource/open-protocols-for-agent-interoperability-part-4-inter-agent-communication-on-a2a/
- Official samples repository: https://github.com/awslabs/amazon-bedrock-agentcore-samples
Top comments (0)