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

AWS details a secure MCP bridge that connects a cloud-hosted Bedrock AgentCore agent to local MCP servers by tunneling signed messages over an existing WebSocket connection.

By AIWI EditorialReviewed by Chayanika DivekarAI-assisted summary · human-reviewed1 min readPublished Aug 5, 2026

Single source1 source

Reality Check

Availability: Availability unclear
Hype-o-meter: Mostly substance
Who it's for: Developers · Enterprises
Do this now: Read the AWS blog post to review the MCP bridge architecture before adapting it to your own AgentCore setup.
Claims check: Claims are limited to the technical approach described; no performance or availability details are stated.
Corroboration: Single source · 1 source

Key Takeaways

Bedrock AgentCore agents run in the cloud while user tools and files stay on local machines
The bridge tunnels signed messages over an existing WebSocket connection
Uses a browser extension and Chrome native messaging to reach local MCP servers
Requires no open ports or VPN

AWS describes how it built a Model Context Protocol (MCP) bridge to let an AI agent hosted on Amazon Bedrock AgentCore access MCP tools running locally on a user’s machine. The challenge addressed is that AgentCore agents operate in the cloud, while users’ tools and files typically reside on their laptops.

The approach tunnels signed messages over an existing WebSocket connection, routing them through a browser extension and Chrome native messaging so the cloud agent can call local MCP servers. According to the post, the setup avoids opening ports or configuring a VPN.

Why it matters

Bridging cloud-hosted agents to local tools and files is a practical requirement for many agent workflows. The described method aims to keep this connection secure while avoiding common networking friction such as open ports and VPNs.

Who should care

Developers building agents on Amazon Bedrock AgentCore, and teams working with MCP-based tooling, may find the described architecture relevant when they need cloud agents to interact with resources on local machines.

Was this helpful?

Sources & Further Reading

How we built an MCP bridge to give our AgentCore-hosted AI agent access to local MCP tools — AWS Machine Learning Blog · Aug 5, 2026

Frequently Asked Questions

How does the bridge connect a cloud agent to local tools?

It tunnels signed messages over the existing WebSocket connection, passing them through a browser extension and Chrome native messaging to reach local MCP servers.

Does it require opening ports or a VPN?

No. According to the description, the bridge works without open ports or a VPN.

Asynchronous patterns for calling Amazon Bedrock AgentCore agents in serverless pipelines — Aug 19, 2026
Monitor on-premises and multi-cloud AI agents with AgentCore Observability — Aug 13, 2026
Run production AI agents in n8n with Amazon Bedrock AgentCore harness — Aug 5, 2026
Authenticate with Private Key JWT using Amazon Bedrock AgentCore Identity — Jul 29, 2026

#agents #apis #cloud #developer-tools