Quickstart

This guide will help you make your first attested TLS connection using Atlas. Choose your platform below to get started.

Node.js
Browser/WASM
Rust

Prerequisites

Node.js 18 or higher
npm or pnpm

Installation

npm install @concrete-security/atlas-node

Basic example

Create a simple attested fetch:

import { createAtlsFetch } from "@concrete-security/atlas-node"

const fetch = createAtlsFetch({
  target: "enclave.example.com",
  policy: {
    type: "dstack_tdx",
    allowed_tcb_status: ["UpToDate", "SWHardeningNeeded"],
    disable_runtime_verification: true  // Development only
  },
  onAttestation: (att) => {
    console.log("TEE Type:", att.teeType)
    console.log("TCB Status:", att.tcbStatus)
    console.log("Trusted:", att.trusted)
  }
})

const response = await fetch("/api/data")
console.log("Status:", response.status)
console.log("Attestation:", response.attestation)

AI SDK integration

Connect to an LLM running in a TEE:

import { createAtlsFetch } from "@concrete-security/atlas-node"
import { createOpenAI } from "@ai-sdk/openai"
import { streamText } from "ai"

const fetch = createAtlsFetch({
  target: "llm.example.com",
  policy: {
    type: "dstack_tdx",
    allowed_tcb_status: ["UpToDate"]
  },
  onAttestation: (att) => console.log(`TEE verified: ${att.teeType}`)
})

const openai = createOpenAI({
  baseURL: "https://llm.example.com/v1",
  apiKey: process.env.OPENAI_API_KEY,
  fetch
})

const { textStream } = await streamText({
  model: openai.chat("your-model"),
  messages: [{ role: "user", content: "Hello from a verified TEE!" }]
})

for await (const chunk of textStream) {
  process.stdout.write(chunk)
}

Use openai.chat(model) for OpenAI-compatible servers like vLLM. The default openai(model) uses the Responses API which most servers don’t support yet.

Next steps

Read the complete Node.js platform guide
Learn about policy configuration
Explore AI SDK integration examples

Prerequisites

Modern browser with WebAssembly support
WebSocket-to-TCP proxy (required for browser connections)

Installation

npm install @concrete-security/atlas-wasm

Setup proxy

Browser applications require a WebSocket-to-TCP proxy:

# Install proxy
cargo install atlas-proxy

# Configure allowlist (required for security)
export ATLS_PROXY_ALLOWLIST="llm.example.com:443"
export ATLS_PROXY_LISTEN="127.0.0.1:9000"

# Start proxy
atlas-proxy

Basic example

import { init, createAtlsFetch } from "@concrete-security/atlas-wasm"

// Initialize WASM module
await init()

const fetch = createAtlsFetch({
  proxyUrl: "ws://127.0.0.1:9000",
  targetHost: "llm.example.com",
  policy: {
    type: "dstack_tdx",
    allowed_tcb_status: ["UpToDate", "SWHardeningNeeded"]
  },
  onAttestation: (att) => {
    console.log("TEE Type:", att.teeType)
    console.log("TCB Status:", att.tcbStatus)
    console.log("Trusted:", att.trusted)
  }
})

const response = await fetch("/v1/chat/completions", {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({
    model: "gpt",
    messages: [{ role: "user", content: "Hello!" }]
  })
})

console.log("Status:", response.status)
console.log("Attestation:", response.attestation)

Streaming responses

Handle streaming LLM responses:

const response = await fetch("/v1/chat/completions", {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({
    model: "gpt",
    messages: [{ role: "user", content: "Hello!" }],
    stream: true
  })
})

const reader = response.body.getReader()
const decoder = new TextDecoder()

while (true) {
  const { done, value } = await reader.read()
  if (done) break
  
  const chunk = decoder.decode(value)
  console.log(chunk)
}

Next steps

Read the complete Browser/WASM platform guide
Learn about proxy setup and configuration
Explore fetch API examples

Prerequisites

Rust 1.75 or higher
Cargo

Installation

Add to your Cargo.toml:

[dependencies]
atlas-rs = "0.2"
tokio = { version = "1", features = ["rt", "macros", "net"] }

Or use cargo add:

cargo add atlas-rs
cargo add tokio --features rt,macros,net

Basic example

Development mode with relaxed verification:

use atlas_rs::{atls_connect, Policy, DstackTdxPolicy};

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let tcp = tokio::net::TcpStream::connect("tee-server.example.com:443").await?;

    // Development policy - relaxed TCB status, no bootchain verification
    let policy = Policy::DstackTdx(DstackTdxPolicy::dev());

    let (mut tls_stream, report) = atls_connect(
        tcp,
        "tee-server.example.com",
        policy,
        None
    ).await?;

    // Access attestation report
    match &report {
        atlas_rs::Report::Tdx(tdx_report) => {
            println!("TEE verified! TCB Status: {}", tdx_report.status);
        }
    }

    // Use tls_stream for subsequent requests...
    Ok(())
}

Full verification (production)

With bootchain measurements and app configuration:

use atlas_rs::{atls_connect, Policy, DstackTdxPolicy, ExpectedBootchain};
use serde_json::json;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let tcp = tokio::net::TcpStream::connect("vllm.example.com:443").await?;

    // Full verification policy
    let policy = Policy::DstackTdx(DstackTdxPolicy {
        expected_bootchain: Some(ExpectedBootchain {
            mrtd: "b24d3b24e9e3c16012376b52362ca09856c4adecb709d5fac33addf1c47e193da075b125b6c364115771390a5461e217".into(),
            rtmr0: "24c15e08c07aa01c531cbd7e8ba28f8cb62e78f6171bf6a8e0800714a65dd5efd3a06bf0cf5433c02bbfac839434b418".into(),
            rtmr1: "6e1afb7464ed0b941e8f5bf5b725cf1df9425e8105e3348dca52502f27c453f3018a28b90749cf05199d5a17820101a7".into(),
            rtmr2: "89e73cedf48f976ffebe8ac1129790ff59a0f52d54d969cb73455b1a79793f1dc16edc3b1fccc0fd65ea5905774bbd57".into(),
        }),
        os_image_hash: Some("86b181377635db21c415f9ece8cc8505f7d4936ad3be7043969005a8c4690c1a".into()),
        app_compose: Some(json!({
            "runner": "docker-compose",
            "docker_compose_file": "version: '3'\nservices:\n  vllm:\n    image: vllm/vllm-openai:latest\n    ..."
        })),
        allowed_tcb_status: vec!["UpToDate".into()],
        ..Default::default()
    });

    let (mut tls_stream, report) = atls_connect(
        tcp,
        "vllm.example.com",
        policy,
        None
    ).await?;

    println!("TEE fully verified!");

    // Use tls_stream for API requests...
    Ok(())
}

JSON policy configuration

Load policies from JSON files:

use atlas_rs::{atls_connect, Policy};

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    // Load policy from JSON
    let policy_json = r#"{
        "type": "dstack_tdx",
        "allowed_tcb_status": ["UpToDate", "OutOfDate"],
        "grace_period": 2592000,
        "expected_bootchain": {
            "mrtd": "b24d3b24e9e3c16012376b52362ca09856c4adecb709d5fac33addf1c47e193da075b125b6c364115771390a5461e217",
            "rtmr0": "24c15e08c07aa01c531cbd7e8ba28f8cb62e78f6171bf6a8e0800714a65dd5efd3a06bf0cf5433c02bbfac839434b418",
            "rtmr1": "6e1afb7464ed0b941e8f5bf5b725cf1df9425e8105e3348dca52502f27c453f3018a28b90749cf05199d5a17820101a7",
            "rtmr2": "89e73cedf48f976ffebe8ac1129790ff59a0f52d54d969cb73455b1a79793f1dc16edc3b1fccc0fd65ea5905774bbd57"
        },
        "os_image_hash": "86b181377635db21c415f9ece8cc8505f7d4936ad3be7043969005a8c4690c1a",
        "app_compose": {
            "runner": "docker-compose",
            "docker_compose_file": "..."
        }
    }"#;

    let policy: Policy = serde_json::from_str(policy_json)?;

    let tcp = tokio::net::TcpStream::connect("tee-server.example.com:443").await?;
    let (tls_stream, report) = atls_connect(
        tcp,
        "tee-server.example.com",
        policy,
        None
    ).await?;

    Ok(())
}

Next steps

Read the complete Rust platform guide
Learn about policy configuration
Explore computing bootchain measurements
Check the Rust API reference

Understanding the examples

All examples above use disable_runtime_verification: true or DstackTdxPolicy::dev() for development. This skips bootchain and application verification but still validates:

The server is running in a genuine TEE
The TCB status meets requirements
The attestation is cryptographically valid

Production deployments must provide full verification with expected_bootchain, os_image_hash, and app_compose. See Policy configuration for details.

Next steps

Platform guides

Complete guides for each platform

Policy configuration

Configure attestation policies for production

Examples

More examples and integration patterns

API reference

Complete API documentation

Get Started

Core Concepts

Policy Configuration

Platform Guides

Deployment

Examples

Prerequisites

Installation

Basic example

AI SDK integration

Next steps

Prerequisites

Installation

Setup proxy

Basic example

Streaming responses

Next steps

Prerequisites

Installation

Basic example

Full verification (production)

JSON policy configuration

Next steps

Understanding the examples

Next steps

Platform guides

Policy configuration

Examples

API reference

Build docs developers (and LLMs) love

Get Started

Core Concepts

Policy Configuration

Platform Guides

Deployment

Examples

​Prerequisites

​Installation

​Basic example

​AI SDK integration

​Next steps

​Prerequisites

​Installation

​Setup proxy

​Basic example

​Streaming responses

​Next steps

​Prerequisites

​Installation

​Basic example

​Full verification (production)

​JSON policy configuration

​Next steps

​Understanding the examples

​Next steps

Platform guides

Policy configuration

Examples

API reference

Build docs developers (and LLMs) love

Prerequisites

Installation

Basic example

AI SDK integration

Next steps

Prerequisites

Installation

Setup proxy

Basic example

Streaming responses

Next steps

Prerequisites

Installation

Basic example

Full verification (production)

JSON policy configuration

Next steps

Understanding the examples

Next steps