Amazon Bedrock AgentCore
Type: full-code · Vendor: Amazon Web Services · Language: Polyglot (managed service + SDKs) · License: proprietary (AWS managed service) · Status: active · Status in practice: emerging
Build, deploy and operate agents securely at scale using any framework and foundation model.
Description. AWS agentic platform of modular, independently-usable services: Runtime (serverless agent execution with session isolation), Harness (a managed agent loop invoked with one API call), Memory, Gateway (turns APIs/Lambda into MCP tools), Identity, Code Interpreter, Browser, Observability (OTEL), Policy (Cedar), Payments (x402), Evaluations and Registry. Framework-agnostic — works with CrewAI, LangGraph, LlamaIndex, Strands, Google ADK and OpenAI Agents SDK — and model-agnostic.
Agent loop shape. Either bring your own framework's loop onto AgentCore Runtime (serverless, isolated microVM sessions), or use AgentCore Harness — a managed agent loop where you specify model, system prompt and tools inline and the platform handles orchestration, tool execution, memory and response generation. Cross-cutting services (Memory, Gateway, Identity, Policy, Observability) attach to either.
Primary use cases
- running any-framework agents serverless with session isolation at scale
- wrapping existing APIs/Lambdas as MCP tools via Gateway
- enterprise governance: identity, policy gating, observability and evaluation for agents
Key concepts
- Harness → react (docs) — Managed agent loop invoked with a single API call.
- Runtime → sandbox-isolation (docs) — Serverless, session-isolated microVM execution for agents.
- Gateway → mcp (docs) — Turns APIs, Lambda and services into MCP-compatible tools.
- Policy → policy-as-code-gate (docs) — Deterministic Cedar/NL rules intercept tool calls before execution.
Patterns this full-code implements —
- ★★ReAct
A managed agent loop that lets you define and invoke AI agents with a single API call. Specify a model, system prompt, and tools inline — Harness handles orc...
- ★★Sandbox Isolation
Each session runs in an isolated microVM with filesystem and shell access
- ★★Model Context Protocol
A secure way to convert your APIs, Lambda functions, and existing services into Model Context Protocol (MCP)-compatible tools
- ★★Cross-Session Memory
Supports for both short-term memory for multi-turn conversations and long-term memory that persists across sessions
- ★Policy-as-Code Gate
Easily author fine-grained rules using natural language or Cedar (AWS's open-source policy language).
- ★Delegated Agent Authorization
A secure, scalable agent identity, access and authentication management service which is compatible with existing identity providers
- ★Agent-Initiated Payment
A fully managed service that enables microtransaction payments for AI agents to access paid APIs, MCP servers, and content using the x402 protocol.
- ★★Eval Harness
A purpose-built evaluation service for automated, consistent, and data-driven agent assessment.
Neighbourhood
Click any neighbour to follow the lineage. Scroll to zoom, drag to pan.
Instantiates
Alternatives & relatives
- full-code · framework
Older AWS managed-agent product; AgentCore is the framework-agnostic platform layer.
- full-code · framework
Azure AI Foundry Agent Service
Microsoft's enterprise agent-runtime platform.
- full-code · framework
Vertex AI Agent Builder
Google Cloud's enterprise agent platform.
- full-code · framework
Strands Agents
AgentCore runs Strands agents as a first-class supported framework.
Listed as alternative by (4)
References
Provenance
- Last analyzed:
- Last updated:
- Verification status: partial