Auditing is a mechanism to capture detailed information about processing. Especially important in highly regulated domains. Auditing is usually scoped to a given instance that can be easily identified e.g., claim, order and so on.

Workflow automation fits very well into this concept as workflow instances are easily identifiable with unique id. By that they can produce a complete audit trail for each and every workflow instance executed.

In many cases, audit information can be very specific and can depend on various factors. Though with workflows it can be categorized to the following:

  • workflow life cycle - starting, completing, aborting or signaling workflow instances

  • workflow node life cycle - triggering workflow nodes (activities and events of the workflow)

  • workflow variables changes - data modifications to variables of given workflow instance

  • timer actions - time based actions that are scheduled and fired on behalf of the workflow instance

  • messaging - messages produced by workflow instances

  • persistence - CRUD operation on the backend data store

These categories can be then included or excluded from being collected as part of the audit trail according to requirements.


Audit mechanism comes with several features out of the box and also allow to extend it in case defaults will not be enough.

First of all, audit is configurable when it comes to categories to include or exclude. Here is a list of categories to be used when configuring:

  • workflow

  • workflow_node

  • workflow_variable

  • workflow_persistence_read

  • workflow_persistence_write

  • timer

  • messaging

These values can be used as comma separated list to either include or exclude given category. Only included and not excluded categories will be stored.

By default all categories are included so if you want to have all audit entries to be stored there is no need to explicitly configure them as included.


Audit entries can be stored pretty much in any data store. By default they are simply logged using logger framework. This comes as a handy default as it allows to to be consumed by centralized logging solutions and by that indexed for further browsing. In addition to that, all features of the logging framework are also available which makes it really flexible.

Though it might not always be enough and to address this potential gap AuditStore is pluggable. To provide your own implementation it is as simple as implementing io.automatiko.engine.api.audit.AuditStore interface in a form of a CDI bean.

It’s recommended to build on top of default implementation (logger based) and extend its capabilities to have still option to store audit entries locally. Default implementation class is io.automatiko.engine.quarkus.audit.LoggerAuditStore


Audit entries can be stored in two formats

  • plain text (default format)

  • JSON

JSON format can be really useful when audit entries are loaded (directly or indirectly) into aggregated logging systems such as `EFK stack' where messages logged as JSON documents can be easily queried by document’s fields.


In some special cases including or excluding categories might not be enough to control what is stored as audit entries. It could be needed to have more control over filtering mechanism.

To make this available, Automatiko Audit comes with an optional filtering mechanism that should be provided by implementors. To build a filter implement io.automatiko.engine.api.audit.AuditEntryFilter interface.

It has a single method accept where a given AuditEntry is given and can be inspected in details. A sample filter implementation could look like the following:

import jakarta.enterprise.context.ApplicationScoped;

import io.automatiko.engine.api.audit.AuditEntry;
import io.automatiko.engine.api.audit.AuditEntryFilter;

public class OrdersOnlyAuditEntryFilter implements AuditEntryFilter {

    public boolean accept(AuditEntry entry) {
        if ("orders".equalsIgnoreCase((String) entry.items().get(AuditEntry.WORKFLOW_DEFINITION_ID))) {
            return true;
        return false;


This filter will only take info account audit entries that have workflowDefinitionId item set to orders

All supported audit entry item names are defined as constants on io.automatiko.engine.api.audit.AuditEntry


Audit mechanism in Automatiko is available in all services out of the box but it is disabled by default.

All categories are included by default, so make sure to configure included or excluded properties based on your needs to avoid too much data to be collected.

Configuration is done in standard way via file. Following are the configuration properties:

Property name Environment variable Description Required Default value



Enables audit logging





Configures format of the audit logging (plain or json)





Types of audit entries that should be logged (workflow,workflow_node,workflow_variable,workflow_persistence_read,workflow_persistence_write,timer,messaging)





Types of audit entries that should be excluded from being logged (workflow,workflow_node,workflow_variable,workflow_persistence_read,workflow_persistence_write,timer,messaging)



Configure audit file

Default implementation of the AuditStore is based on logger, due to that it allows to use pretty much all the features of the logging system. One of useful things is to store all audit entries into a separate file. To do so configure category logger for AutomatikoAudit as follows:


What this does is:

  • created named (AUDIT_FILE) file logger handler

  • sets a logging format to only store message (and new line after the message)

  • configure INFO level for AutomatikoAudit category

  • configures file location - target/audit.log

  • enables this logger handler for AutomatikoAudit category

  • disables parent handler to avoid duplicated writes (to file and into console) of audit events

There could be more properties to define rotation of the file.