Architect runtime helpers are optional but they do make working with CloudFormation provisioned resources nicer. CloudFormation resources are generated with names more friendly for machines than people. Other frameworks leave resource discovery up to end users which leads to ad hoc implementations becoming a frequent bug vector. Architect treats runtime discovery as a first class concern.

Amazon Resource Names (ARNs) are available at runtime to all Lambda functions defined in the same app.arc. Things such as DynamoDB tables, SNS topics, SQS queues, API Gateway endpoints, and S3 static bucket ARNs are baked into @architect/functions so your runtime program logic interacts with resources using people friendly and readable names defined in the app.arc file.


Install the Architect runtime helpers for Node:

npm install @architect/functions

Ensure arc is available to your Lambda function code:

let arc = require('@architect/functions')



Get a static asset path:

let css = arc.static('/index.css')


Middleware with async functions is defined on arc.http.async with middleware functions as parameters. The returned function adheres to the expected AWS Lambda function signature. A function can exit the middleware queue early by returning an HTTP response.

let arc = require('@architect/functions')

exports.handler = arc.http.async(auth, handler)

async function auth(request) {
  if (!request.session.account) {
    return { status: 403 }

async function handler(request) {
  return {
    json: { ok: true }


The incoming request object is the standard API Gateway request with a few enhancements:

  • body is automatically parsed
  • session automatically parsed from the request cookie


Architect honors the standard API Gateway response payload parameters:

  • statusCode
  • headers
  • body
  • isBase64Encoded

And adds the following clean convenience params:

  • cacheControl sets the Cache-Control header
  • css sets the Content-Type header to text/css; charset=utf8
  • code alias for statusCode
  • cors sets the Access-Control-Allow-Origin header to *
  • html sets the Content-Type header to text/html; charset=utf8
  • js sets the Content-Type header to text/javascript; charset=utf8
  • json sets the Content-Type header to application/json
  • session write a value to the session
  • status also an alias for statusCode
  • text sets the Content-Type header to text/plain; charset=utf8
  • type sets the Content-Type header
  • xml sets the Content-Type header to text/xml; charset=utf8

Express migration helper.

let arc = require('@architect/functions')
let express = require('express')

let app = express()

app.get('/', (req, res) => res.send('Hello World!'))
app.get('/cool', (req, res)=> res.send('very cool'))

exports.handler =


Middleware for serving @static folder assets.

let arc = require('@architect/functions')

let asap = arc.http.proxy({
  spa: false,
  alias: {
    '/playground': '/playground.html'

exports.handler = arc.http.async(asap)


Read the current session in an @http request and write it back to a cookie.

async function handler (req) {
  // read the session
  let session = await
  // save the session into a cookie string
  let cookie = await arc.http.session.write({ count: 1 })
  // write the cookie to the browser
  return { 
    statusCode: 200,
    headers: { 'set-cookie': cookie },


Create a DynamoDB client for @tables.

Given the following app.arc file:


  personID *String
  noteID **String

Generate a data access layer:

let arc = require('@architect/functions')
let data = await arc.tables()

For the example above the generated API is:

  • data.notes.get
  • data.notes.query
  • data.notes.scan
  • data.notes.put
  • data.notes.delete
  • data.notes.update

The generated client is facade for AWS.DynamoDB.DocumentClient. The delete and get methods take a single parameter that is passed on to the params.Key attribute in the corresponding DocumentClient method. The put method takes a single parameter that is passed on as the params.Item attribute in the DocumentClient.put method. The query, scan, and update methods simply pass the params argument with the TableName parameter prepopulated. See the official DynamoDB documentation for all available parameters.

The generated data layer also allows direct access to DynamoDB through a few methods:

  • data._db which returns an instance of AWS.DynamoDB
  • data._doc returns an instance of AWS.DynamoDB.DocumentClient
  • data._name helper function that returns a @table resource name when you need to go lower level. For example use data._name("my-table") to get the name of the "my-table" @table resource.

Subscribe to an SNS topic

let arc = require('@architect/functions')

exports.handler =

async function handler (event) {

Publish to an SNS topic

let arc = require('@architect/functions')

  name: 'hit-counter',
  payload: {hits: 1},


Subscribe to an SQS queue

let arc = require('@architect/functions')

exports.handler = arc.queues.subscribe(handler)

async function handler (event) {

Publish to an SQS queue

let arc = require('@architect/functions')

await arc.queues.publish({
  name: 'hit-counter',
  payload: {hits: 1},

Send a message over WebSockets.

let arc = require('@architect/functions')

  id: connectionId,
  payload: { message: 'hai 🐶' }