arc sandbox

Architect projects work locally and offline. Sandbox emulates most app resources defined in app.arc:

  • @http
  • @static
  • @ws
  • @events
  • @queues
  • @tables and @tables-indexes

Additionally, @scheduled and @tables-streams Lambdas can be emulated via the @architect/plugin-lambda-invoker plugin


arc sandbox [--port|--host|--disable-symlinks|--no-hydrate|--verbose]


  • -p, --port - Manually specify HTTP port
    • Defaults to 3333
  • -h, --host - Specify the host interface for Sandbox to listen on
    • Defaults to (all available interfaces on your machine)
    • To accept local connections only, specify localhost
  • -v, --verbose - Enable verbose logging
  • -d, --debug - Enable debug logging
  • -q, --quiet - Disable (most) logging
  • --disable-symlinks - Disable symlinking src/shared into all functions and use file copying instead

CLI variables

The following variables can be set on the command line when running arc sandbox. Other variables will be ignored by Sandbox.

  • ARC_API_TYPE - Set the API Gateway API type
    • Can be one of http (aliased to httpv2), httpv1, rest
    • Defaults to http
  • ARC_ENV - testing|staging|production
    • Defaults to testing
  • ARC_HOST - Specify the host interface for Sandbox to listen on
    • Defaults to (all available interfaces on your machine)
    • To accept local connections only, specify localhost
  • ARC_LOCAL- If present and used in conjunction with ARC_ENV=staging|production, emulates live staging or production environment
    • Uses your local preferences @env environment variables for the appropriate stage
    • Connects Sandbox to live AWS events and DynamoDB infrastructure
    • Requires valid AWS credentials with the same profile name as defined in your project manifest
  • Specify ports:
    • ARC_PORT - Manually specify HTTP port
      • Defaults to 3333
    • ARC_EVENTS_PORT- Manually specify event bus port
      • Defaults to 4444
    • ARC_TABLES_PORT- Manually specify local DynamoDB port
      • Defaults to 5555
    • ARC_INTERNAL_PORT- Manually specify internal Sandbox + AWS services port
      • Defaults to 2222
  • ARC_DB_EXTERNAL - (Boolean) Use an external DynamoDB tool (such as AWS NoSQL Workbench)
  • ARC_QUIET - If present, disable (most) logging


Run Sandbox in quiet mode on a different port:

ARC_QUIET=1 PORT=8888 npx arc sandbox

Local preferences

Several Architect local preferences can be leveraged to change how Sandbox works while developing locally.


The following can be set as a part of the @sandbox pragma.

  • livereload - Enable live automatic reload for @http get and any functions that deliver HTML. When a filesystem change is detected in the handler or in shared or views code, open browser sessions will automatically refresh.
    • Defaults to false
  • env* - Override the local environment setting to use staging or production so that Sandbox uses that stage’s environment variables as set in local preferences @env or in the project’s .env file.
    • Can be one of testing (default), staging, or production
  • useAWS* - Use live AWS infrastructure from Sandbox. Specifically, @tables, @tables-indexes, @events, and @queues.
    • Uses the staging environment by default, but env can be set to production.
    • Defaults to false
  • no-hydrate - Disable function hydration on Sandbox start.
    • Defaults to false
  • seed-data - Specify a custom file path for seed data to populate @tables with on startup
    • Defaults to ./sandbox-seed.json, ./sandbox-seed.js
  • external-db - (Boolean) Use an external DynamoDB tool (such as AWS NoSQL Workbench)
livereload true
env production
useAws true
no-hydrate true

* These advanced options should be used with care since they will allow local development code to interact with live AWS resources.


Additionally, Sandbox can run shell commands on startup by setting @sandbox-start in local preferences like so:

node scripts/seed_db.js
echo 'hello'


Upon starting, Sandbox can automatically scaffold resources (via arc init) found in the application’s manifest that do not yet exist. Options are set with @create in local preferences.

  • autocreate - Set to true to enable automatic creation of boilerplate Lambda handlers and static assets if they do not exist.
  • templates - Specify templates for automatic resource scaffolding.
    • <pragma name> path/to/template.ext
    • Does not enable autocreate
autocreate true
  http path/to/template/http.js
  events path/to/template/


Architect Sandbox will load variables for Sandbox’s current environment (testing, staging, or production) from a local preferences file with @env. If a project contains a .env file, Architect will load those variables instead, and only for the testing environment.

Variables from local preference files and .env will not be merged. Further details, including the variable load-strategy are outlined below.

Environment variables

Sandbox automatically loads environment variables for availability at runtime (process.env.MY_VAR in Node.js). Environment variables can be set in a few locations. It’s important to understand how each source is prioritized when developing locally.

Load strategy

Sandbox will prioritize…

  1. A project’s .env file (if it exists),
  2. then project-level Architect preferences,
  3. and finally global Architect preferences.

Variables across these sources are not merged.

Using a local preferences file with @env offers the most flexibility since variables can be specified per environment: testing, staging, and production.

Example scenario

If .env is found, Sandbox will only use the variables for the testing environment, and not load any variables from any Arc preferences files. Given the following case with 3 environment variable sources:

# ./.env
# ./prefs.arc
  URL www.architect.fake
# ~/.preferences.arc
  URL ftp://arc.ftp
  ADMIN_PASS zero cool

The following is true in a Node.js function run with Sandbox:

process.env.URL === '' // true
process.env.ADMIN_PASS // undefined

Local database

Sandbox creates an in-memory instance of dynalite with @tables and @tables-indexes found in the app.arc file. @tables-streams is not currently supported by dynalite.

When Sandbox is terminated, any data written is cleared from memory.

You can set a custom port by using an environment variable: ARC_TABLES_PORT=5555.

Database seed data

You can automatically seed data to Sandbox upon startup by adding a sandbox-seed.js or sandbox-seed.json file to the root of your project. (You can also specify a custom path with the seed-data preference.)

Your seed data should be an object whose properties correspond to @tables names, and have arrays of rows to seed. For example:

  id *String
  sort **String
// sandbox-seed.js
module.exports = {
  things: [
      id: 'foo',
      sort: 'bar',
      arbitrary: 'data',
      id: 'fiz',
      sort: 'buz',
      arbitrary: 'info',

The above example would add the two rows above to the things database each time Sandbox is started.

Note: This feature is only enabled if the environment is testing, so as to prevent the accidental (over)writing of data to a live database.

Live database example

Connect Sandbox to the DynamoDB staging database on AWS:

ARC_ENV=staging ARC_LOCAL=1 npx arc sandbox