A Ruby client for the Amazon Ads API.

Add to your Gemfile:

gem "amazon-ads"

Or install directly:

gem install amazon-ads

Set your app credentials at the module level or via environment variables:

AmazonAds.client_id = "your_client_id"
AmazonAds.client_secret = "your_client_secret"

Or set AMAZON_ADS_CLIENT_ID and AMAZON_ADS_CLIENT_SECRET.

Request an access token via Login with Amazon (LWA):

data = AmazonAds::LWA.request(refresh_token: "your_refresh_token")
access_token = data.fetch("access_token")

The caller owns token caching. Store and reuse the token until it expires.

Make requests:

# List advertising profiles
profiles = AmazonAds::Profiles.new(region: "NA", access_token:)
profiles.list_profiles

# List campaigns under a profile
campaigns = AmazonAds::Campaigns.new(
  region: "NA",
  access_token:,
  profile_id: "123456789",
)
campaigns.list_campaigns

Pass http: to inject a configured HTTP client (e.g. http: HTTP.use(logging: { logger: Logger.new($stdout) })). Note that debug logging prints request headers, exposing access tokens.

Responses with a 4xx or 5xx status raise AmazonAds::Error, which carries the full response:

begin
  campaigns.list_campaigns
rescue AmazonAds::Error => e
  e.status         # 400
  e.code           # "FIELD_VALUE_IS_INVALID"
  e.response.body  # Amazon's error document
end

When the response carries one of Amazon's documented error codes, the error is a matching subclass under AmazonAds::Errors:

begin
  campaigns.list_campaigns
rescue AmazonAds::Errors::TooManyRequests
  backoff
rescue AmazonAds::Errors::Unauthorized
  refresh_token
end

Unknown or missing codes raise the base AmazonAds::Error. Reporting sends the HTTP status as its code, so its errors are always the base class.

The error supports pattern matching on status and code:

case error
in code: "TOO_MANY_REQUESTS" then backoff
in status: 500..599 then retry
end

Network failures (HTTP::ConnectionError, HTTP::TimeoutError) raise as-is, whether or not retries are configured.

See AGENTS.md.