AWS IAM policy statement generator with fluent interface for AWS CDK
Project description
IAM Floyd
AWS IAM policy statement generator with fluent interface.
Support for:
- 239 Services
- 7951 Actions
- 781 Resource Types
- 458 Conditions
This is an early version of the package. The API will change while I implement new features. Therefore make sure you use an exact version in your package.json
before it reaches 1.0.0.
Packages
There are two different package variants available:
- iam-floyd: Can be used in AWS SDK, Boto 3 or for whatever you need an IAM policy statement for
- cdk-iam-floyd: Integrates into AWS CDK and extends
iam.PolicyStatement
Usage
Depending on your scenario, you need to either import iam-floyd
or cdk-iam-floyd
:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
# for use without AWS CDK use the iam-floyd package
import iam_floyd as statement
# for use with CDK use the cdk-iam-floyd package
import cdk_iam_floyd as statement
Both packages contain a statement provider for each AWS service, e.g. Ec2
. A statement provider is a class with methods for each and every available action, resource type and condition. Calling such method will add the action/resource/condition to the statement:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().to_start_instances()
Every method returns the statement provider, so you can chain method calls:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().to_start_instances().to_stop_instances()
The default effect of any statement is Allow
. To add some linguistic sugar you can explicitly call the allow()
method:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().allow().to_start_instances().to_stop_instances()
And of course deny()
:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().deny().to_start_instances().to_stop_instances()
If you don't want to be verbose and add every single action manually to the statement, you can work with access levels. For every access level there is a distinct method available to add all related actions to the statement:
allListActions()
allReadActions()
allWriteActions()
allPermissionManagementActions()
allTaggingActions()
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().deny().all_permission_management_actions()
statement.Ec2().allow().all_list_actions().all_read_actions()
To add actions based on regular expressions, use the method allMatchingActions()
.
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().deny().all_matching_actions("/vpn/i")
To add all actions (e.g. ec2:*
), call the allActions()
method:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().allow().all_actions()
For every available condition key, there are if*()
methods available.
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().allow().to_start_instances().if_encrypted().if_instance_type(["t3.micro", "t3.nano"]).if_associate_public_ip_address(False).if_aws_request_tag("Owner", "John")
If you want to add a condition not covered by the available methods, you can define just any condition yourself via if()
:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().allow().to_start_instances().if("aws:RequestTag/Owner", "John")
The default operator for conditions of type String is StringLike
.
Most of the if*()
methods allow an optional operator as last argument:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().allow().to_start_instances().if("aws:RequestTag/Owner", "*John*", "StringEquals")
Statements without principals, by default, apply to all resources. To limit to specific resources, add them via on*()
.
For every resource type an on*()
method exists:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.S3().allow().all_actions().on_bucket("some-bucket").on_object("some-bucket", "some/path/*")
If instead you have an ARN ready, use the on()
method:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.S3().allow().all_actions().on("arn:aws:s3:::some-bucket", "arn:aws:s3:::another-bucket")
To invert the policy you can use notActions()
and notResources()
:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.S3().allow().not_actions().not_resources().to_delete_bucket().on_bucket("some-bucket")
Examples
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
policy = {
"Version": "2012-10-17",
"Statement": [
statement.Ec2().allow().to_start_instances().if_aws_request_tag("Owner", "${aws:username}"),
statement.Ec2().allow().to_stop_instances().if_resource_tag("Owner", "${aws:username}"),
statement.Ec2().allow().all_list_actions().all_read_actions()
]
}
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
policy = {
"Version": "2012-10-17",
"Statement": [
statement.Cloudformation().allow().all_actions(),
statement.All().allow().all_actions().if_aws_called_via("cloudformation.amazonaws.com"),
statement.S3().allow().all_actions().on("arn:aws:s3:::cdktoolkit-stagingbucket-*"),
statement.Account().deny().all_permission_management_actions().all_write_actions(),
statement.Organizations().deny().all_permission_management_actions().all_write_actions()
]
}
Methods
allow
Sets the Effect
of the statement to Allow
.
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().allow().to_stop_instances()
deny
Sets the Effect
of the statement to Deny
.
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().deny().to_stop_instances()
to*, to
For every available action, there are to*()
methods available.
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().allow().to_start_instances().to_stop_instances()
allActions
This method adds all actions of the related service to the statement, e.g. ec2:*
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().allow().all_actions()
allMatchingActions
Adds all actions matching regular expressions to the statement.
The regular expressions need to be in Perl/JavaScript literal style and need to be passed as strings:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().allow().all_matching_actions("/vpn/i")
allListActions
Adds all actions with access level list to the statement.
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().allow().all_list_actions()
allReadActions
Adds all actions with access level read to the statement.
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().allow().all_read_actions()
allWriteActions
Adds all actions with access level write to the statement.
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().allow().all_write_actions()
allPermissionManagementActions
Adds all actions with access level permission management to the statement.
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().allow().all_permission_management_actions()
allTaggingActions
Adds all actions with access level tagging to the statement.
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().allow().all_tagging_actions()
if*, if
For every available condition key, there are if*()
methods available.
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().allow().to_start_instances().if_encrypted().if_instance_type(["t3.micro", "t3.nano"]).if_associate_public_ip_address(False).if_aws_request_tag("Owner", "John")
Most of them allow an optional operator as last argument:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().allow().to_start_instances().if_instance_type("*.nano", "StringLike")
Global conditions are prefixed with ifAws
, e.g. ifAwsRequestedRegion()
If you want to add a condition not covered by the available methods, you can define just any condition yourself via if()
:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Ec2().allow().to_start_instances().if("aws:RequestTag/Owner", "${aws:username}", "StringEquals")
on*, on
Limit statement to specified resources.
For every resource type an on*()
method exists:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.S3().allow().all_actions().on_bucket("some-bucket")
If instead you have an ARN ready, use the on()
method:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.S3().allow().all_actions().on("arn:aws:s3:::some-bucket")
If no resources are applied to the statement, it defaults to all resources (*
). You can also be verbose and set this yourself:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.S3().allow().all_actions().on("*")
notActions
Switches the policy provider to use NotAction.
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.S3().allow().not_actions().to_delete_bucket().on_bucket("some-bucket")
notResources
Switches the policy provider to use NotResource.
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.S3().allow().not_resources().to_delete_bucket().on_bucket("some-bucket")
notPrincipals
Switches the policy provider to use NotPrincipal.
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Sts().deny().not_principals().to_assume_role().for_user("1234567890", "Bob")
for*
To create assume policies, use the for*()
methods. There are methods available for any type of principal:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Sts().allow().to_assume_role().for_account("1234567890")
statement.Sts().allow().to_assume_role_with_sAML().for_service("lambda.amazonaws.com")
statement.Sts().allow().to_assume_role().for_user("1234567890", "Bob")
statement.Sts().allow().to_assume_role().for_role("1234567890", "role-name")
statement.Sts().allow().to_assume_role_with_sAML().for_federated_cognito()
statement.Sts().allow().to_assume_role_with_sAML().for_federated_amazon()
statement.Sts().allow().to_assume_role_with_sAML().for_federated_google()
statement.Sts().allow().to_assume_role_with_sAML().for_federated_facebook()
statement.Sts().allow().to_assume_role_with_sAML().for_saml("1234567890", "saml-provider")
statement.Sts().allow().to_assume_role().for_public()
statement.Sts().allow().to_assume_role().for_assumed_role_session("123456789", "role-name", "session-name")
statement.Sts().allow().to_assume_role().for_canonical_user("userID")
statement.Sts().allow().to_assume_role().for("arn:foo:bar")
To reverse the assume policy you can call the notPrincipals()
method:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Sts().deny().not_principals().to_assume_role().for_user("1234567890", "Bob")
If you use the cdk variant of the package you should not have the need to manually create assume policies. But if you do, there is an additional method forCdkPrincipal()
which takes any number of iam.IPrincipal
objects:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.Sts().allow().to_assume_role().for_cdk_principal(
iam.ServicePrincipal("sns.amazonaws.com"),
iam.ServicePrincipal("lambda.amazonaws.com"))
Collections
The package provides commonly used statement collections. These can be called via new statement.Collection().allowEc2InstanceDeleteByOwner()
. Collections return a list of statements, which then can be used in a policy like this:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
policy = {
"Version": "2012-10-17",
"Statement": [
(SpreadElement ...new statement.Collection().allowEc2InstanceDeleteByOwner()
statement.Collection().allow_ec2_instance_delete_by_owner())
]
}
In Python you would need to convert the statements to JSON first:
import iam_floyd as statement
statements = statement.Collection().allow_ec2_instance_delete_by_owner()
policy = {
'Version': '2012-10-17',
'Statement': list(map(lambda x: x.to_json(), statements)),
}
Available collections are:
- allowEc2InstanceDeleteByOwner: Allows stopping EC2 instance only for the user who started them
allowEc2InstanceDeleteByOwner
Allows stopping EC2 instance only for the user who started them.
FAQ
Why should I use this package instead of writing policies by hand?
All actions, conditions and resource types of every service are explorable via code suggestion. The related documentation is available in the method description. In most cases you can avoid reading the documentation completely.
IntelliSense makes it super easy to find what you're looking for. But it also helps with discovering things you were not looking for! Users write more secure/restrictive policies because they can easily type .if
and add conditions with a <tab>
without looking up multiple documentation pages.
By calling methods of a class you protect yourself against typos. If your code doesn't compile/run because of a typo, you'll immediately notice. If instead you have a typo in your action list, IAM will silently accept your policy. You won't notice until you see a warning in the IAM console.
Allowing/Denying all actions based on access level is a feature AWS missed when designing IAM policies. With this package it is as easy as calling .allWriteActions()
, .allReadActions()
etc.
In IAM policies you can use wildcards to add actions to the statement. Wildcards often do not have enough power to define patterns and quickly include too many actions. This package enables you to select actions with regular expressions.
Limiting actions to specific resources via ARN is cumbersome. In this package, for every resource type there is a method, which not only helps with ARN creation - it also adds context to the code which helps to understand the meaning. The classical example here is to allow all actions on an S3 bucket and its containing objects:
{
"Effect": "Allow",
"Action": "s3:*",
"Resource": [
"arn:aws:s3:::example-bucket"
"arn:aws:s3:::example-bucket/*"
]
}
The first resource element is for the bucket itself. The second element is for the contained objects.
A beginner might make the mistake to think the first entry is superfluous and remove it. This package has distinct methods to limit actions to a bucket and/or objects:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
statement.S3().allow().all_actions().on_bucket("example-bucket").on_object("example-bucket", "*")
And yes, it's shorter too.
How often will there be updates to reflect IAM changes?
Once per hour the AWS documentation is checked for updates. If anything changes, a new package will be released immediately.
Do you release new packages when a new CDK version is released?
No. I believe it's a myth and a user error if packages are incompatible with new releases of the CDK. cdk-iam-floyd
is based on cdk ^1.30.0
and so far I have not seen any issues.
Is the package following semantic versioning?
Mostly. For manual changes by developers this package follows semver.
Automatic releases triggered by changes in the IAM documentation will always result in a minor update.
It has been observed that IAM actions have been deleted or renamed. This case will not be reflected by a major update! If you had been using such a method your code will break. On the other hand, your code probably already is broken, since it creates a policy with invalid actions until you update to the latest release.
I don't like method chaining!
That's not a question. But yes, you can completely avoid method chaining:
# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
my_statement = statement.Ec2()
my_statement.allow()
my_statement.to_start_instances()
my_statement.to_stop_instances()
Floyd?
George Floyd has been murdered by racist police officers on May 25, 2020.
This package is not named after him to just remind you of him and his death. I want this package to be of great help to you and I want you to use it on a daily base. Every time you use it, I want you to remember our society is ill and needs change. The riots will stop. The news will fade. The issue persists!
If this statement annoys you, this package is not for you.
Similar projects
Legal
The code contained in the lib/generated folder is generated from the AWS documentation. The class- and function-names and their description therefore are property of AWS.
AWS and their services are trademarks, registered trademarks or trade dress of AWS in the U.S. and/or other countries.
This project is not affiliated, funded, or in any way associated with AWS.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for cdk_iam_floyd-0.71.0-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | ff49f9fa97d73b8f3ec02c7bdd35690a0f3ea794865bd0601591e80ad3f3760d |
|
MD5 | 7c527fc6cd0d25f954d9aa3b93d57eaa |
|
BLAKE2b-256 | 2757b8a5be4450bb77022af5a209335b8ea570106511a937916b026700d430bb |