Architect Inventory

All Architect plugins receive an inventory object that enumerates all the resources in a project. So what is the Architect Inventory format, anyway?

How Architect Inventory works

Every time Architect initiates a command, your project manifest is parsed and interpreted into the internal Architect Inventory format that serves as the single source of truth for your project’s state. Whenever your project resources or configuration changes in any way, it is reflected in the Inventory.

For example, let’s say your project manifest looks like this:

get /foo

Inventory responsible for enumerating the HTTP pragma’s get /foo route, and its current configuration (e.g. Node.js version, memory, timeout, etc.) based on Architect’s defaults, any project-level defaults you may have set, and any per-Lambda configuration via config.arc.

Inventory is also responsible for lower-level project configuration. For example: the above project does not define a handler at the root of your API; Inventory automatically creates a configuration to ensure something exists to handle requests to your root (so your users don’t get a 5xx error when visiting the root of your site).

In the same vein, Inventory would also set up default project options for static assets in anticipation of their use by your HTTP route(s).

The Inventory format

The Inventory format represents every possible setting, option, and configuration in Architect. Since it isn’t practical to document all that here. We’ll cover the basics to help you navigate it.

The basics:

  • Plugins are passed both arc (the raw manifest in the parsed Architect format) and inventory (the Architect Inventory format, probably what you’ll most often be using)
    • The arc format is still helpful to have for inspecting custom pragmas. For example, if you wanted your users to configure your plugin with a custom @my-plugin-config pragma, that would only appear in arc, and not in inventory
  • The arc + inventory objects passed to plugins are frozen copies; only Architect is responsible for changing or updating current inventory
  • inventory contains two objects:
    • inv: the current project Inventory
    • get: a convenient set of resource getter methods
  • With limited exceptions, inventory values of null represent the lack of a userland configuration; anything non-null almost always represents something that was user-configured (or inferred by configuration)


The inv object contains a property for each Architect pragma, and a handful of meta properties and resource collections. Here is a basic example inv object:

inventory.inv = {
  _arc:             { /* Architect metadata: current version, pragmas, stage, etc. */ },
  _project:         { /* Project metadata: cwd, preferences, env vars, etc. */ },
  ...pragmas,         /* Properties for each pragma: aws, http, tables, etc. */,
  lambdaSrcDirs:    [ /* Array of the source directory of each project Lambda */ ],
  lambdasBySrcDir:  { /* Object containing each Lambda, named by its source directory */ },

Each pragma’s Lambdas may contain different properties and configurations; for example, @scheduled Lambdas have a rate or cron property, while @http Lambdas have a method and path property.

Here’s an example of what the above get /foo Lambda object might look like:

inventory.inv.http = [
    name: 'get /foo',
    method: 'get',
    path: '/foo',
    // Lambda configuration
    config: {
      timeout: 10,
      memory: 3008,
      runtime: 'nodejs20.x',
      architecture: 'arm64',
      handler: 'index.handler',
      state: 'n/a',
      concurrency: 'unthrottled',
      layers: [],
      policies: [],
      shared: true,
      env: true,
    // Other Architect config + paths and files
    src: '/your/project/path/src',
    handlerFile: '/your/project/path/src/http/get-foo/index.js',
    handlerMethod: 'handler',
    handlerModuleSystem: 'esm',
    configFile: '/your/project/path/src/http/get-foo/config.arc',
    pragma: 'http'

We encourage you to explore the Inventory format on your own; a quick way to inspect your project’s Inventory would be to just log it out from a plugin like so:

module.exports = { sandbox: { start: function ({ inventory }) {
  console.dir(inventory.inv, { depth: null })
} } }

Note: remember, set plugins run before the rest of the Inventory, and thus receive incomplete Inventory objects.


The get object contains a variety of helpful methods for querying the inventory. For example, say a project had the following manifest:

get /foo
post /bar
get /fiz
put /buz

fingerprint true

Now say you wanted to get the Inventory Lambda object of @http get /foo. You could use find() on the http array to search for the name property of get /foo (e.g. inventory.inv.http.find(({ name }) => name === 'get /foo')). Or you can use inventory.get:

console.log(get.http('get /foo')) // Your `get /foo` Lambda
console.log(get.http('get /bar')) // null

get also works on settings pragmas, like @static or @aws:

console.log(get.static('fingerprint'))  // true
console.log(get.static('prune'))        // null
console.log('region'))          // 'us-west-2'