Need help with skew?
Click the “chat” button below for chat support from the developer who created it, or find similar developers for support.

About the developer

229 Stars 68 Forks Apache License 2.0 296 Commits 28 Opened issues

Services available


Need anything else?

Contributors list



is a package for identifying and enumerating cloud resources. The name is a homonym for SKU (Stock Keeping Unit). Skew allows you to define different SKU
which are a particular encoding of a SKU. Skew then allows you to use this scheme pattern and regular expressions based on the scheme pattern to identify and enumerate a resource or set of resources.

At the moment, the the only available

is the
scheme. The
scheme uses the basic structure of Amazon Resource Names (ARNs) to assign a unique identifier to every AWS resource.

An example ARN pattern would be:


This pattern identifies a specific EC2 instance running in the

region under the account ID
. The account ID is the 12-digit unique identifier for a specific AWS account as described here. To allow
to find your account number, you need to create a
YAML config file. By default,
will look for your config file in
but you can use the
environment variable to tell
where to find your config file if you choose to put it somewhere else. The basic format of the
config file is:
      profile: dev
      profile: prod

Within the

section, you create keys named after your 12-digit account ID (as a string). Within that, you must have an entry called profile that lists the profile name this account maps to within your AWS credential file.

The main purpose of skew is to identify resources or sets of resources across services, regions, and accounts and to quickly and easily return the data associated with those resources. For example, if you wanted to return the data associated with the example ARN above:

from skew import scan

arn = scan('arn:aws:ec2:us-west-2:123456789012:instance/i-12345678') for resource in arn: print(

The call to

returns an ARN object which implements the iterator pattern and returns a
object for each AWS resource that matches the ARN pattern provided. The
object contains all of the data associated with the AWS resource in dictionary under the

Any of the elements of the ARN can be replaced with a regular expression. The simplest regular expression is

which means all available choices. So, for example:
arn = scan('arn:aws:ec2:us-east-1:*:instance/*')

would return an iterator for all EC2 instances in the

region found in all accounts defined in the config file.

To find all DynamoDB tables in all US regions for the account ID 234567890123 you would use:

arn = scan('arn:aws:dynamodb:us-.*:234567890123:table/*')

CloudWatch Metrics

In addition to making the metadata about a particular AWS resource available to you,

also tries to make it easy to access the available CloudWatch metrics for a given resource.

For example, assume that you had did a

on the original ARN above and had the resource associated with that instance available as the variable
. You could do the following:
>>> instance.metric_names


attribute returns the list of available CloudWatch metrics for this resource. The retrieve the metric data for one of these:
>>> instance.get_metric_data('CPUUtilization')
[{'Average': 0.134, 'Timestamp': '2014-09-29T14:04:00Z', 'Unit': 'Percent'},
 {'Average': 0.066, 'Timestamp': '2014-09-29T13:54:00Z', 'Unit': 'Percent'},
 {'Average': 0.066, 'Timestamp': '2014-09-29T14:09:00Z', 'Unit': 'Percent'},
 {'Average': 0.134, 'Timestamp': '2014-09-29T13:34:00Z', 'Unit': 'Percent'},
 {'Average': 0.066, 'Timestamp': '2014-09-29T14:19:00Z', 'Unit': 'Percent'},
 {'Average': 0.068, 'Timestamp': '2014-09-29T13:44:00Z', 'Unit': 'Percent'},
 {'Average': 0.134, 'Timestamp': '2014-09-29T14:14:00Z', 'Unit': 'Percent'},
 {'Average': 0.066, 'Timestamp': '2014-09-29T13:29:00Z', 'Unit': 'Percent'},
 {'Average': 0.132, 'Timestamp': '2014-09-29T13:59:00Z', 'Unit': 'Percent'},
 {'Average': 0.134, 'Timestamp': '2014-09-29T13:49:00Z', 'Unit': 'Percent'},
 {'Average': 0.134, 'Timestamp': '2014-09-29T13:39:00Z', 'Unit': 'Percent'}]

You can also customize the data returned rather than using the default settings:

>>> instance.get_metric_data('CPUUtilization', hours=8, statistics=['Average', 'Minimum', 'Maximum'])
[{'Average': 0.132,
  'Maximum': 0.33,
  'Minimum': 0.0,
  'Timestamp': '2014-09-29T10:54:00Z',
  'Unit': 'Percent'},
 {'Average': 0.134,
  'Maximum': 0.34,
  'Minimum': 0.0,
  'Timestamp': '2014-09-29T14:04:00Z',
  'Unit': 'Percent'},
 {'Average': 0.066,
  'Maximum': 0.33,
  'Minimum': 0.0,
  'Timestamp': '2014-09-29T08:34:00Z',
  'Unit': 'Percent'},
 {'Average': 0.134,
  'Maximum': 0.34,
  'Minimum': 0.0,
  'Timestamp': '2014-09-29T08:04:00Z',
  'Unit': 'Percent'}]

Filtering Data

Each resource that is retrieved is a Python dictionary. Some of these (e.g. an EC2 Instance) can be quite large and complex. Skew allows you to filter the data returned by applying a jmespath query to the resulting data. If you aren't familiar with jmespath, check it out. Its a very powerful query language for JSON data and has full support in Python as well as a number of other languages such as Ruby, PHP, and Javascript. It is also the query language used in the AWSCLI so if you are familiar with the

option there, you can use the same thing with skew.

To specify a query to be applied to results of a scan, simply append the query to the end of the ARN, separated by a

(pipe) character. For example:

Would retrieve the data for this particular EC2 instance and would then filter the returned data through the (very) simple jmespath query to which retrieves the value of the attribute

within the data. The filtered data is available as the
attribute of the Resource object. The full, unfiltered data is still available as the

Multithreaded Usage

Skew is single-threaded by default, like most Python libraries. In order to speed up the enumeration of matching resources, you can use multiple threads:

import skew

class Worker(Thread): def init(self, arn): Thread.init(self) self.arn = arn = arn

def run(self): for i in skew.scan(self.arn): # now do something with the stuff

arn = skew.ARN()

for service in arn.service.choices(): uri = 'arn:aws:' + service + ':::/' worker = Worker(uri); worker.start()

(thanks to @alFReD-NSH for the snippet)

More Examples

Find Unattached Volumes

Audit Security Groups

Find Untagged Instances

We use cookies. If you continue to browse the site, you agree to the use of cookies. For more information on our use of cookies please see our Privacy Policy.