Configuration Using Local Policy File (Docker)

For deployments of open-appsec on Docker the configuration can be done in a declarative way using a single YAML file which holds all the relevant configuration objects. This way it is fully compatible for integration in GitOps CD-based processes.

Alternatively management can also be done centrally via the central open-appsec WebUI (SaaS).

Location of the declarative configuration file

On Docker the declarative configuration file for the open-appsec Agent is typically provided to the agent container via a volume mount.

In your docker-compose.yaml or the docker run command which you used for your current deployment you can find the specific folder on the docker host which gets mounted to /ext/appsec in the open-appsec agent container. This is the folder which should contain your local_policy.yaml file. (Note that this folder then gets mounted internally to /etc/cp/policy/local_policy.yaml which is where the agent looks for the actual local policy to apply, but on Docker you should always manage the file in /ext/appsec file location)

You can show and edit the full default declarative configuration file with the following command:

From within the appsec-agent container:

vi /ext/appsec/local_policy.yaml

or by using the interactive CLI tool open-appsec-ctl : Using the open-appsec-ctl Tool

From the docker host:

(replace "appsec-agent" with your own agent container name, if you adjusted it):

docker exec -it appsec-agent vi /ext/appsec/local_policy.yaml

Basic configuration

The default policy within the default configuration file, which is deployed during the installation, contains the following setting which sets the mode to detect-learn for all web resources provided by the NGINX or Kong Gateway:

policies.default.mode: detect-learn

If you want attacks instead to be prevented by default for all web resources you can change this as follows:

policies.default.mode: prevent-learn

The specific rules section policies.specific-rules (v1beta1) / policies.specificRules (v1beta2) allows you to create specific rule entries for specific hostnames, hostname-path combinations or paths which override the default policy.

The mode settings defined in the policy for default or specific-rules will only have an effect on the actual enforcement for specific security features when the override-mode setting in the referenced practice is set to as-top-level as this will cause it to inherited the mode setting from the policy.

It is recommended to configure a specific rule for each of your protected web applications and web APIs assets (starting with detect-learn mode, later move to prevent-learn). This allows you to apply a separate enforcement mode setting for each of these protected assets as well as to customize many other settings like selecting the logging configuration, the threat prevention and access control (introduced in v1beta2) practices configuration or selecting a custom-response like a specific response code or a custom block page.

Also the real-time contextual machine learning will be done individually per each configured specific rule representing a protected asset. Note that these settings are all configured by referencing other elements which are defined separately in the configuration file as well, so that you can reuse them multiple times in a kind of object-oriented way, see examples below:

v1beta1

policies:
  default:
  ...
  ...
  specific-rules:
  - host: "www.my-specific-host.com/my-specific-path"
    triggers:
    - appsec-default-log-trigger
    mode: detect-learn
    practices:
    - webapp-default-practice
    custom-response: appsec-default-web-user-response

It is recommended that once sufficient confidence was gained in detect-learn mode for an asset which has a specific rule (review the logs) to change the mode for this specific rule to prevent-learn mode:

policies:
  default:
  ...
  ...
  specific-rules:
  - host: "www.my-specific-host.com/my-specific-path"
    triggers:
    - appsec-default-log-trigger
    mode: prevent-learn
    practices:
    - webapp-default-practice
    custom-response: appsec-default-web-user-response

v1beta2

policies:
  default:
  ...
  ...
  specificRules:
  - host: "www.my-specific-host.com/my-specific-path"
    triggers:
    - appsec-default-log-trigger
    mode: detect-learn
    threatPreventionPractices:
    - webapp-default-practice
    accessControlPractices:
    - webapp-default-practice
    customResponse: appsec-default-web-user-response

It is recommended that once sufficient confidence was gained in detect-learn mode for an asset which has a specific rule (review the logs) to change the mode for this specific rule to prevent-learn mode:

policies:
  default:
  ...
  ...
  specificRules:
  - host: "www.my-specific-host.com/my-specific-path"
    triggers:
    - appsec-default-log-trigger
    mode: prevent-learn
    threatPreventionPractices:
    - webapp-default-practice
    accessControlPractices:
    - webapp-default-practice
    customResponse: appsec-default-web-user-response

Applying a changed local declarative configuration file

Once you did any changes to the local declarative configuration file local_policy.yaml there's two ways how you can get those changes applied:

1. By using the open-appsec-ctl tool to apply the changes in the policy:

• From inside the Agent container:

open-appsec-ctl --apply-policy

• From the docker host:

docker exec -it appsec-agent open-appsec-ctl --apply-policy

Replace appsec-agent with your agent container name, if it has a different name.

1. By configuring the agent to automatically apply all changes in the local_policy.yaml file in /ext/appsec once you save any changes within the file: Activate automatic policy load for the open-appsec agent by setting the following environment variable to true in your docker-compose.yaml or docker run command: autoPolicyLoad=true This is especially useful in CI/CD-style deployments that have a continuous deployment process in place.