AWS Bedrock AgentCore — Claude Code MCP server: what it does
AWS Bedrock AgentCore is an MCP server that connects AI agents to Amazon Bedrock AgentCore APIs for runtime, memory, identity, gateway, and policy operations through 122 tools.
Apache-2.0 ★ 9,939 updated 2026-10-23
Open on GitHubWhat AWS Bedrock AgentCore does
The server exposes seven AgentCore primitives: Runtime (14 tools) deploys agent runtimes and endpoints, Memory (21) handles events and semantic search, Identity (21) manages OAuth2 providers and token vaults, Gateway (15) creates API gateways and resource policies, Policy (15) operates policy engines, Browser (25) automates cloud-based browsing, Code Interpreter (10) runs sandboxed code execution. Documentation tools (search_agentcore_docs, fetch_agentcore_doc) work independently.
Calls real boto3 APIs. Requires configured AWS credentials (AWS_PROFILE, AWS_ACCESS_KEY_ID, or IAM role) and Python 3.10+. Without credentials, resource creation and management tools will not start.
All primitives are enabled by default. AGENTCORE_DISABLE_TOOLS turns off specific ones (e.g., browser,code_interpreter), AGENTCORE_ENABLE_TOOLS registers only the listed primitives. Documentation tools remain available regardless.
What it helps with
- deploy an agent runtime in AgentCore from Claude Code
- store an event in agent memory and run semantic search
- configure an OAuth2 identity provider and issue API keys
- create an API gateway and attach a resource policy
- execute code in a sandboxed environment with file upload
Example prompts for the agent
- Create an agent runtime in
us-east-1with default settings - Store this event in memory with tags
user-feedbackandbug-report - Search AgentCore docs for gateway access policy configuration
How to install AWS Bedrock AgentCore
Install
The MCP registry has no launch command for it: see the repository README
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about AWS Bedrock AgentCore
How to install AWS Bedrock AgentCore in Claude Code?
Install uv from Astral, then Python 3.10+ with uv python install 3.10. Add the server to your MCP config (~/.kiro/settings/mcp.json) with command uvx and argument awslabs.amazon-bedrock-agentcore-mcp-server@latest. Configure AWS credentials via AWS_PROFILE or environment variables.
What does AWS Bedrock AgentCore do?
Connects AI agents to Amazon Bedrock AgentCore operational APIs: runtime and endpoint management, memory and semantic search, OAuth2 providers and API keys, API gateways, policy engines, cloud browser automation, sandboxed code execution. 122 tools call real boto3 APIs.
What does AWS Bedrock AgentCore need to work?
Python 3.10+, the uv utility, and configured AWS credentials (profile, keys, or IAM role). Without credentials, resource creation and management tools will not start.