Home icon

How we built an MCP bridge to give our AgentCore-hosted AI agent access to local MCP tools

Machine Learning Blog

This article describes building an MCP bridge that enables cloud-hosted AI agents on Amazon Bedrock AgentCore to access local MCP tools through a browser extension and native messaging.

  • MCP bridge tunnels messages between remote AgentCore agent and local MCP servers using WebSocket and native messaging
  • Browser extension acts as bidirectional relay, forwarding MCP JSON-RPC messages between cloud and local machine
  • FastMCP proxy translates between native messaging protocol and standard MCP JSON-RPC over stdio
  • Presigned WebSocket URLs with SigV4 signing keep credentials on user's machine, expiring after 5 minutes
  • MCP servers configured via mcp.json file; new tools available automatically without agent code changes
  • Production recommendations include JWT authentication, payload signing, file system scoping, and audit logging

The architecture enables centrally deployed AI agents to securely access user's local files and tools while maintaining process isolation and credential security.

Go to article

The AWS News Feed is currently looking for gold sponsors. If you want to support the AWS community and reach a large audience of AWS professionals, consider sponsoring the AWS News Feed.

Related articles

Jul 28
2026
How AgentCore Gateway supports the MCP 2026-07-28 spec
Dec 8
2025
Streamline AI agent tool interactions: Connect API Gateway to AgentCore Gateway with MCP
Nov 6
2025
Transform your MCP architecture: Unite MCP servers through AgentCore Gateway
Jun 1
2026
Extending MCP support for Amazon Bedrock AgentCore Gateway

The AWS News Feed is currently looking for silver sponsors. If you want to support the AWS community and reach a large audience of AWS professionals, consider sponsoring the AWS News Feed.