attestable-mcp-server
Enables remote attestation of MCP servers using Intel SGX and RA-TLS, allowing clients to verify the exact code running on the server.
README
➡️ attestable-mcp-server
<div align="center">
<strong>remotely attestable MCP server</strong> </div>
Overview
This project contains an MCP Server that is remotely attestable by MCP clients. To achieve this, a trusted execution environment is used, which generates a certificate representing the currently-running code of the attestable-mcp-server. The attestable-mcp-server sends this certificate in the TLS handshake to an MCP client before connecting that proves the code it's running is the same code built on github actions, and can be independently validated by building and running the code locally on emulated hardware or secure hardware; these values will be the same. The protocol used for client <-> server remote attestation is RA-TLS, an extension to TLS that adds machine and code specific measurements that can be verified by an MCP client.
The most important concept behind this RA-TLS certificate is that it embeds an SGX quote in the standardized X.509 extension field with the TCG DICE "tagged evidence" OID, which in turn embeds the SGX report and the complete Intel SGX certificate chain. In addition to the SGX quote, the certificate also contains the evidence claims, with the most important one being the "pubkey-hash" claim that contains the hash of the ephemeral public key (in DER format) generated by the TEE of the memory image of the running MCP server.
<strong>Features</strong>
- MCP Clients can remotely attest the code running on any MCP Server
- MCP Servers can optionally remotely attest MCP Clients
Producing Signed Artifacts
The github action script in this repo runs on a self-hosted github runner inside of a trusted execution environment (TEE). The action script will build a docker container containing the attestable-mcp-server and generate a signed attestation of the code running inside the TEE. This docker image is then signed by github. You can independently generate the same values with or without secure hardware, and query our running server and get the same values.
Dependencies
- Intel SGX Hardware
- Gramine
- python 3.13
- Ubuntu 22.04
- Intel SGX SDK & PSW
Quickstart
uv sync
docker build -t attestable-mcp-server .
gramine-sgx-gen-private-key
git clone https://github.com/gramineproject/gsc docker/gsc
cd docker/gsc
uv run ./gsc build-gramine --rm --no-cache -c ../gramine_base.config.yaml gramine_base
uv run ./gsc build -c ../attestable-mcp-server.config.yaml --rm attestable-mcp-server ../attestable-mcp-server.manifest
uv run ./gsc sign-image -c ../attestable-mcp-server.config.yaml attestable-mcp-server "$HOME"/.config/gramine/enclave-key.pem
uv run ./gsc info-image gsc-attestable-mcp-server
Starting Server on Secure Hardware
docker run -itp --device=/dev/sgx_provision:/dev/sgx/provision --device=/dev/sgx_enclave:/dev/sgx/enclave -v /var/run/aesmd/aesm.socket:/var/run/aesmd/aesm.socket -p 8000:8000 --rm gsc-attestable-mcp-server
Starting Server on local development machine
docker run -p 8000:8000 --rm gsc-attestable-mcp-server
TODO
- add MCP client demonstrating ra-tls
- add intel-signed measurements from our github action to this readme for simple independent verification
Future Plans
- JSON Web Key (JWK) attestation claim validation
cobrowser.xyz
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.