Skip to content
forked from aws/aws-cli

Universal Command Line Interface for Amazon Web Services


Notifications You must be signed in to change notification settings



Repository files navigation


Build Status Gitter

This package provides a unified command line interface to Amazon Web Services.

The aws-cli package works on Python versions:

  • 2.6.5 and greater
  • 2.7.x and greater
  • 3.3.x and greater
  • 3.4.x and greater
  • 3.5.x and greater
  • 3.6.x and greater
  • 3.7.x and greater


We recommend that all customers regularly monitor the Amazon Web Services Security Bulletins website for any important security bulletins related to aws-cli.


The easiest way to install aws-cli is to use pip in a virtualenv:

$ pip install awscli

or, if you are not installing in a virtualenv, to install globally:

$ sudo pip install awscli

or for your user:

$ pip install --user awscli

If you have the aws-cli installed and want to upgrade to the latest version you can run:

$ pip install --upgrade awscli


On OS X, if you see an error regarding the version of six that came with distutils in El Capitan, use the --ignore-installed option:

$ sudo pip install awscli --ignore-installed six

This will install the aws-cli package as well as all dependencies. You can also just download the tarball. Once you have the awscli directory structure on your workstation, you can just run:

$ cd <path_to_awscli>
$ python install

If you want to run the develop branch of the CLI, see the "CLI Dev Version" section below.

CLI Releases

The release notes for the AWS CLI can be found here.

Command Completion

The aws-cli package includes a very useful command completion feature. This feature is not automatically installed so you need to configure it manually. To enable tab completion for bash either use the built-in command complete:

$ complete -C aws_completer aws

Or add bin/aws_bash_completer file under /etc/bash_completion.d, /usr/local/etc/bash_completion.d or any other bash_completion.d location.

For tcsh:

$ complete aws 'p/*/`aws_completer`/'

You should add this to your startup scripts to enable it for future sessions.

For zsh please refer to bin/ Source that file, e.g. from your ~/.zshrc, and make sure you run compinit before:

$ source bin/

For now the bash compatibility auto completion (bashcompinit) is used. For further details please refer to the top of bin/

Getting Started

Before using aws-cli, you need to tell it about your AWS credentials. You can do this in several ways:

  • Environment variables
  • Shared credentials file
  • Config file
  • IAM Role

The quickest way to get started is to run the aws configure command:

$ aws configure
AWS Access Key ID: foo
AWS Secret Access Key: bar
Default region name [us-west-2]: us-west-2
Default output format [None]: json

To use environment variables, do the following:

$ export AWS_ACCESS_KEY_ID=<access_key>
$ export AWS_SECRET_ACCESS_KEY=<secret_key>

To use the shared credentials file, create an INI formatted file like this:



and place it in ~/.aws/credentials (or in %UserProfile%\.aws/credentials on Windows). If you wish to place the shared credentials file in a different location than the one specified above, you need to tell aws-cli where to find it. Do this by setting the appropriate environment variable:

$ export AWS_SHARED_CREDENTIALS_FILE=/path/to/shared_credentials_file

To use a config file, create a configuration file like this:

aws_access_key_id=<default access key>
aws_secret_access_key=<default secret key>
# Optional, to define default region for this profile.

[profile testing]
aws_access_key_id=<testing access key>
aws_secret_access_key=<testing secret key>

and place it in ~/.aws/config (or in %UserProfile%\.aws\config on Windows). If you wish to place the config file in a different location than the one specified above, you need to tell aws-cli where to find it. Do this by setting the appropriate environment variable:

$ export AWS_CONFIG_FILE=/path/to/config_file

As you can see, you can have multiple profiles defined in both the shared credentials file and the configuration file. You can then specify which profile to use by using the --profile option. If no profile is specified the default profile is used.

In the config file, except for the default profile, you must prefix each config section of a profile group with profile. For example, if you have a profile named "testing" the section header would be [profile testing].

The final option for credentials is highly recommended if you are using aws-cli on an EC2 instance. IAM Roles are a great way to have credentials installed automatically on your instance. If you are using IAM Roles, aws-cli will find them and use them automatically.

Other Configurable Variables

In addition to credentials, a number of other variables can be configured either with environment variables, configuration file entries or both. The following table documents these.

Variable Option Config Entry Environment Variable Description
profile --profile profile AWS_PROFILE Default profile name
region --region region AWS_DEFAULT_REGION Default AWS Region
config_file     AWS_CONFIG_FILE Alternate location of config
credentials_file     AWS_SHARED_CREDENTIALS_FILE Alternate location of credentials
output --output output AWS_DEFAULT_OUTPUT Default output style
ca_bundle --ca-bundle ca_bundle AWS_CA_BUNDLE CA Certificate Bundle
access_key   aws_access_key_id AWS_ACCESS_KEY_ID AWS Access Key
secret_key   aws_secret_access_key AWS_SECRET_ACCESS_KEY AWS Secret Key
token   aws_session_token AWS_SESSION_TOKEN AWS Token (temp credentials)
cli_timestamp_format   cli_timestamp_format   Output format of timestamps
metadata_service_timeout   metadata_service_timeout AWS_METADATA_SERVICE_TIMEOUT EC2 metadata timeout
metadata_service_num_attempts   metadata_service_num_attempts AWS_METADATA_SERVICE_NUM_ATTEMPTS EC2 metadata retry count
parameter_validation   parameter_validation   Toggles local parameter validation


If you get tired of specifying a --region option on the command line all of the time, you can specify a default region to use whenever no explicit --region option is included using the region variable. To specify this using an environment variable:

$ export AWS_DEFAULT_REGION=us-west-2

To include it in your config file:

aws_access_key_id=<default access key>
aws_secret_access_key=<default secret key>

Similarly, the profile variable can be used to specify which profile to use if one is not explicitly specified on the command line via the --profile option. To set this via environment variable:

$ export AWS_PROFILE=testing

The profile variable can not be specified in the configuration file since it would have to be associated with a profile and would defeat the purpose.

Further Information

For more information about configuration options, please refer the AWS CLI Configuration Variables topic. You can access this topic from the CLI as well by running aws help config-vars.

Accessing Services With Global Endpoints

Some services, such as AWS Identity and Access Management (IAM) have a single, global endpoint rather than different endpoints for each region.

To make access to these services simpler, aws-cli will automatically use the global endpoint unless you explicitly supply a region (using the --region option) or a profile (using the --profile option). Therefore, the following:

$ aws iam list-users

will automatically use the global endpoint for the IAM service regardless of the value of the AWS_DEFAULT_REGION environment variable or the region variable specified in your profile.

JSON Parameter Input

Many options that need to be provided are simple string or numeric values. However, some operations require JSON data structures as input parameters either on the command line or in files.

For example, consider the command to authorize access to an EC2 security group. In this case, we will add ingress access to port 22 for all IP addresses:

$ aws ec2 authorize-security-group-ingress --group-name MySecurityGroup \
  --ip-permissions '{"FromPort":22,"ToPort":22,"IpProtocol":"tcp","IpRanges":[{"CidrIp": ""}]}'

File-based Parameter Input

Some parameter values are so large or so complex that it would be easier to place the parameter value in a file and refer to that file rather than entering the value directly on the command line.

Let's use the authorize-security-group-ingress command shown above. Rather than provide the value of the --ip-permissions parameter directly in the command, you could first store the values in a file. Let's call the file ip_perms.json:


Then, we could make the same call as above like this:

$ aws ec2 authorize-security-group-ingress --group-name MySecurityGroup \
    --ip-permissions file://ip_perms.json

The file:// prefix on the parameter value signals that the parameter value is actually a reference to a file that contains the actual parameter value. aws-cli will open the file, read the value and use that value as the parameter value.

This is also useful when the parameter is really referring to file-based data. For example, the --user-data option of the aws ec2 run-instances command or the --public-key-material parameter of the aws ec2 import-key-pair command.

URI-based Parameter Input

Similar to the file-based input described above, aws-cli also includes a way to use data from a URI as the value of a parameter. The idea is exactly the same except the prefix used is https:// or http://:

$ aws ec2 authorize-security-group-ingress --group-name MySecurityGroup \

Command Output

The default output for commands is currently JSON. You can use the --query option to extract the output elements from this JSON document. For more information on the expression language used for the --query argument, you can read the JMESPath Tutorial.


Get a list of IAM user names:

$ aws iam list-users --query Users[].UserName

Get a list of key names and their sizes in an S3 bucket:

$ aws s3api list-objects --bucket b --query Contents[].[Key,Size]

Get a list of all EC2 instances and include their Instance ID, State Name, and their Name (if they've been tagged with a Name):

$ aws ec2 describe-instances --query \
  'Reservations[].Instances[].[InstanceId,State.Name,Tags[?Key==`Name`] | [0].Value]'

You may also find the jq tool useful in processing the JSON output for other uses.

There is also an ASCII table format available. You can select this style with the --output table option or you can make this style your default output style via environment variable or config file entry as described above. Try adding --output table to the above commands.

CLI Dev Version

If you are just interested in using the latest released version of the AWS CLI, please see the Installation section above. This section is for anyone who wants to install the development version of the CLI. You normally would not need to do this unless:

  • You are developing a feature for the CLI and plan on submitting a Pull Request.
  • You want to test the latest changes of the CLI before they make it into an official release.

The latest changes to the CLI are in the develop branch on github. This is the default branch when you clone the git repository.

Additionally, there are several other packages that are developed in lockstep with the CLI. This includes:

If you just want to install a snapshot of the latest development version of the CLI, you can use the requirements.txt file included in this repo. This file points to the development version of the above packages:

cd <path_to_awscli>
pip install -r requirements.txt
pip install -e .

However, to keep up to date, you will continually have to run the pip install -r requirements.txt file to pull in the latest changes from the develop branches of botocore, jmespath, etc.

You can optionally clone each of those repositories and run "pip install -e ." for each repository:

git clone <jmespath> && cd jmespath/
pip install -e . && cd ..
git clone <botocore> && cd botocore/
pip install -e . && cd ..
git clone <awscli> && cd aws-cli/
pip install -e .

Getting Help

We use GitHub issues for tracking bugs and feature requests and have limited bandwidth to address them. Please use these community resources for getting help:


Universal Command Line Interface for Amazon Web Services







No packages published


  • Python 99.9%
  • Other 0.1%