Trusted GMail MCP Server

Trusted GMail MCP Server

First Trusted MCP server running on AWS Nitro Enclave Trusted Execution Environment

0xfreysa

Cloud Platforms
Visit Server

README

Trusted GMail MCP Server

This is a gmail MCP server running inside a secure AWS Nitro enclave instance. It was originally forked from the Claude Post MCP server. Most MCP servers are run locally via the stdio transport; we followed this guide to implement a remote MCP server using sse transport.

Connect to the MCP Server

To use this MCP server, you will need an app-specific password.

Then simply add the following block to your client's mcp.json file.

    "gmail_mcp": {
      "url": "https://gmail.mcp.freysa.ai/sse/?ADDR=<your.email@gmail.com>&ASP=<your app-specific password>"
    }

Note that you might have to restart your client.

Security Notice

This implementation is a proof of concept. Passing app-specific passwords in URLs is not a secure pattern because:

  • URLs can be logged by proxies, browsers and servers
  • URLs may appear in browser history
  • URLs can be leaked via the Referer header to third-party sites

Unfortunately, current MCP clients have limitations on how they connect to servers. At the moment of release, MCP specification does not define a standard authentication mechanism for SSE servers. This means we can't use more secure patterns like bearer tokens or other authorization headers that would normally be preferred.

For additional security, consider:

  1. Using a dedicated app-specific password just for this purpose
  2. Accessing this over a secure VPN or private network
  3. Running your own instance with the provided instructions

Concept

AWS Nitro Enclaves provide isolated compute environments that enhance security through hardware-based attestation. When code runs in a Nitro Enclave, the platform generates cryptographic measurements of the code's identity and state. These measurements serve as a verifiable guarantee that the code has not been modified and is executing exactly as intended, protecting against tampering or unauthorized modifications. For more information, see this blog post.

We use Nitriding to quickly deploy code in an AWS Nitro TEE.

Verify the code attestation

To verify that the intended codebase is the one running in our TEE, you must reproduce running it in an AWS Nitro enclave yourself. Instructions to do so are below. Once you have it running, you can verify it using this repository as follows.

  1. First build the code.
cd verifier

pnpm install && pnpm run build
  1. Then run the verifier locally.
cd mcp/react-ts-webpack

pnpm i && pnpm run dev
  1. Then open http://localhost:8080/ in your browser. You will be prompted to add two fields

(a) the PCR2 hash, which is a hash of the codebase

(b) the Code attestation, which is signed by AWS

  1. Click the "Verify Attestation" button

Run your own instance in a TEE

You can reproduce running this server in a TEE as follows.

  1. Use the AWS EC2 console to select a sufficiently large instance and be sure to enable Nitro.

  2. Make sure that the ports needed by your application are open by checking the security group, in "security" tab of the instance in the ec2 console.

  3. Clone this repo to your ec2 instance.

  4. Run the setup script to download all necessary dependencies.

sudo /setup.sh
  1. Allocate more memory for the enclave if necessary.
sudo nano /etc/nitro_enclaves/allocator.yaml

sudo systemctl restart nitro-enclaves-allocator.service
  1. Run the enclave.
make
  1. Run in production mode.
make run

Use your MCP server

To actually use the MCP server, you will also need to run the gvproxy, as follows.

screen
./gvproxy.sh

Then you can curl the healthcheck endpoint to confirm that the MCP server is running in the enclave.

curl http://127.0.0.1:7047/

Recommended Servers

contentful-mcp

contentful-mcp

Update, create, delete content, content-models and assets in your Contentful Space

Featured
TypeScript
Supabase MCP Server

Supabase MCP Server

A Model Context Protocol (MCP) server that provides programmatic access to the Supabase Management API. This server allows AI models and other clients to manage Supabase projects and organizations through a standardized interface.

Featured
JavaScript
Azure MCP Server

Azure MCP Server

Enables natural language interaction with Azure services through Claude Desktop, supporting resource management, subscription handling, and tenant selection with secure authentication.

Official
Local
TypeScript
SettleMint

SettleMint

Leverage SettleMint's Model Context Protocol server to seamlessly interact with enterprise blockchain infrastructure. Build, deploy, and manage smart contracts through AI-powered assistants, streamlining your blockchain development workflow for maximum efficiency.

Official
Local
TypeScript
Brev

Brev

Run, build, train, and deploy ML models on the cloud.

Official
Local
Python
Story SDK MCP Server

Story SDK MCP Server

This server provides MCP (Model Context Protocol) tools for interacting with Story's Python SDK. Features Get license terms Mint and register IP Asset with PIL Terms Mint license tokens Send $IP to a wallet Upload image to ipfs via Pinata [External] Upload ip and nft metadata via Pinata [External]

Official
Python
Tembo MCP Server

Tembo MCP Server

An MCP server that enables Claude to interact with Tembo Cloud platform API, allowing users to manage Tembo Cloud resources through natural language.

Official
TypeScript
Workers MCP

Workers MCP

A package that connects Claude Desktop and other MCP clients to Cloudflare Workers, enabling custom functionality to be accessed via natural language through the Model Context Protocol.

Official
TypeScript
Appwrite MCP Server

Appwrite MCP Server

A Model Context Protocol server that allows AI assistants to interact with Appwrite's API, providing tools to manage databases, users, functions, teams, and other resources within Appwrite projects.

Official
Python
MCP2Lambda

MCP2Lambda

Enables AI models to interact with AWS Lambda functions via the MCP protocol, allowing access to private resources, real-time data, and custom computation in a secure environment.

Official
Python