In short
Activiti is an open-source, lightweight workflow and business process management (BPM) platform whose core is a BPMN 2.0 process engine for Java. Teams model processes as BPMN files and deploy them to the engine. Script tasks inside those files run with broad access to the JVM, so attackers who can deploy a model may try to run code.
Key takeaways
- What it is
- An Apache-2.0-licensed Java workflow and BPM platform. Its core engine runs processes described in BPMN 2.0 XML.
- How it works
- Process definitions are deployed to the engine, often through a REST API. Script tasks run code in any JSR-223 language, such as JavaScript or Groovy, as part of the process.
- Why it matters
- The user guide warns that a script task has similar capabilities to a Java delegate, with full access to the JVM. A deployed model can therefore act like code.
- What Odin Research saw
- An Activiti workbench of 81 files in Monograph 05, with BPMN models built to evade web application firewalls and to call Groovy. No operating-system command on a target was shown.
- What to do first
- Keep model import and REST endpoints off the internet, disable unrestricted Groovy and JSR-223 scripts, and review every BPMN import like a code release.
What is Activiti?
Activiti is an open-source workflow and business process management (BPM) platform whose core is a BPMN 2.0 process engine for Java. Companies use it to automate approvals, orders and other business flows. Process models can contain script tasks that run code inside the engine, which makes model deployment a security-sensitive action.
The GitHub README calls Activiti a light-weight workflow and BPM platform aimed at business users, developers and system admins. It is licensed under Apache 2.0, integrates with Spring and had about 10,600 stars and 6,900 forks on 21 November 2026. A related project, Activiti Cloud, builds on the same engine. Odin Research's reports refer to it as Apache Activiti.
Activiti is ordinary enterprise middleware, not a hacking tool. It matters to defenders because, like other workflow engines, it compiles and runs custom logic by design. Monograph 05 notes that workstation security tools rarely watch such engines.
How Activiti works
A process is drawn or written in BPMN 2.0, an XML standard for business processes. It contains events, gateways and tasks. The engine runs each process instance step by step and stores its state.
- Deployment. Process definitions are deployed to the engine. The Activiti 6.0.0 user guide documents REST operations to list, create and delete deployments and their resources.
- Service tasks. A service task calls Java code or evaluates an expression.
- Script tasks. A script task carries code inside the process file. The guide says its
scriptFormatattribute must name a JSR-223 scripting engine. JavaScript comes with the JDK, and Groovy works once its library is on the classpath. All process variables are visible to the script.
That last point is the risk. The guide's Secure Scripting section says that, by default, a script task has similar capabilities to a Java delegate. It has full access to the JVM, can run forever and can use a lot of memory. Unlike Java code, scripts ship inside the process definition and can run as soon as it is deployed. The guide adds that scripts are sometimes not written by developers.
Why Activiti matters to defenders
An attacker who can deploy a BPMN model to an Activiti engine may not need a classic exploit. A script task or expression does the work under the engine's service account. Monograph 05 sums it up: an attacker needs a valid session or an exposed model-import endpoint on an engine that allows scripts.
That shifts the defense:
- Model imports are code deployments. They deserve the same review, signing and change control.
- Engine credentials are powerful. Workflow engines often hold service credentials for the systems they automate, so a foothold spreads. See least privilege.
- WAFs can be dodged. A code execution payload does not have to contain obvious shell strings, as the next section shows.
In ATT&CK terms, this sits between Exploit Public-Facing Application (T1190) and Command and Scripting Interpreter (T1059).
What Odin Research observed
Monograph 05 studied staged toolkits aimed at enterprise middleware: Epicor Kinetic, SQL Server Agent, VMware Cloud Director and the n8n and Activiti workflow engines. One dedicated server held the Activiti toolkit.
- 81 files, seen on 13 November 2026. They included 35 JSON and 18 XML files, 13 of them named as BPMN 2.0 definitions with
rcein the name. - Stage 1: no shell strings. A builder script carried a Chinese comment that translates as a plan to use pure JVM code without shell signature strings to evade the WAF. Its model had one script node that read the JVM user name and host name and sent them back over HTTP.
- Stage 2: Groovy and Spring EL. One model defined a Groovy
scriptTaskthat would create aProcessBuilder. Another used a service-task expression that called the engine's scripting engines to send an HTTP callback. - Saved replies. The operator kept REST replies from an Activiti endpoint, whose address is withheld. They show test deployments accepted and one test process completed after evaluating the harmless expression
${1+1}. - Staged files, not outcomes. No target-side record showed an operating-system command run through a workflow engine.
How to detect Activiti abuse
- Script content in deployments. Flag deployment or model-save requests whose body contains
groovytogether withProcessBuilder, orgetScriptingEngines().evaluate, in plain text or Base64. - Recon-only models. Script nodes that only read system properties or the local host name, then open a URL, fit the stage-1 pattern.
- Child processes. A Java workflow runtime, such as Activiti or Flowable, that starts a shell or
curlis a strong signal. Record process creation with full command lines on workflow servers, for example withauditdEXECVE rules on Linux. - Unexpected deployments. Compare the engine's deployment list with your release records. A test process you did not ship deserves review.
Monograph 05 publishes Suricata signatures for plaintext and Base64 Groovy and Spring EL uploads and stage-1 Modeler uploads, a Sigma rule for workflow runtimes starting shells, and a YARA rule for Activiti and n8n payloads. Find them in the Defense Kit.
How to secure Activiti
- Hide the engine. Keep the REST API and any model-import endpoint off the internet and behind strong authentication. See How to respond to misconfigurations.
- Restrict scripting. Monograph 05 advises disabling unrestricted Groovy and JSR-223
scriptTaskexecution. The user guide documents an experimental Secure Scripting feature, added in Activiti 5.21, and a section on enabling safe BPMN 2.0 XML. - Review every import. Require role-based review of every BPMN 2.0 XML and Modeler JSON import, as you would for application code.
- Least privilege for the engine. Run it under a dedicated account with only the database and network access it needs.
- Patch the stack. Keep the engine, Spring and scripting libraries current. See Vulnerability Management 101.
Activiti compared with other middleware in Monograph 05
| Platform | What the staged toolkit targeted | Key defense from the report |
|---|---|---|
| Activiti | BPMN 2.0 models with Groovy script tasks and Spring EL expressions, built to evade WAFs | Disable unrestricted Groovy; review every BPMN import |
| n8n | Session forgery and Code Node injection calling child_process | Sandboxed task runners without access to environment variables |
| Epicor Kinetic | Business-rule directive injection through database tables | Audit directive table writes |
| SQL Server Agent | Job steps that run operating-system commands | Restrict CmdExec steps |
| VMware Cloud Director Availability | Path traversal and root password reset tooling | Keep management ports off the internet |
Frequently asked questions about Activiti
What is Activiti?
Activiti is an open-source, Apache-2.0-licensed workflow and business process management platform. Its core is a BPMN 2.0 process engine for Java that runs business processes defined in XML.
Is Activiti malware?
No. Activiti is legitimate enterprise software. Its script tasks can run code inside the engine, so attackers who can deploy process models may try to abuse it, as staged files in Odin Research's Monograph 05 show.
What is an Activiti script task?
A script task is a BPMN step that runs a script inside the engine. The script language must be JSR-223 compatible, such as JavaScript or Groovy, and the script can read all process variables.
Is Activiti secure?
It can be run securely, but the user guide warns that a script task by default has full access to the JVM. Keep the engine private, restrict scripting and review every deployed model.
Is Activiti an Apache project?
Activiti is licensed under the Apache License 2.0, and Odin Research's reports refer to it as Apache Activiti. Its code is hosted on GitHub under the Activiti organization.
How can attackers abuse Activiti?
By deploying a BPMN model whose script task or expression runs their code under the engine's account. They need a valid session or an exposed model-import endpoint on an engine that allows scripts.
How do I detect Activiti abuse?
Inspect deployment requests for Groovy with ProcessBuilder or scripting-engine calls, compare deployed models with release records, and alert when the Java workflow runtime starts a shell. Odin Research publishes Suricata, Sigma and YARA rules for this.
Activiti vs n8n: what is the difference?
Activiti is a Java BPMN 2.0 engine for business processes. n8n is a workflow automation tool with JavaScript Code Nodes. Both run custom code by design, and Monograph 05 found staged toolkits aimed at each.
Sources
- Activiti repository and README, GitHub. github.com/Activiti/Activiti (checked 21 November 2026)
- Activiti 6.0.0 User Guide (script task, Secure Scripting and REST API sections). www.activiti.org/userguide/ (checked 21 November 2026)
Related tools and terms
What is Epicor Kinetic?
Epicor Kinetic is an enterprise resource planning (ERP) suite for manufacturers and distributors. Its business process management (BPM) directives can run custom code inside the ERP.
Enterprise softwareWhat is VMware Cloud Director?
VMware Cloud Director is a multi-tenant cloud management platform that service providers use to run virtual data centres for customers.
Enterprise softwareWhat is VMware ESXi?
VMware ESXi is a bare-metal hypervisor that runs virtual machines. Ransomware groups target it because one host can hold many servers.
Enterprise softwareWhat is n8n?
n8n is an open-source workflow automation platform. Its workflows can run code and hold credentials for connected services.
Enterprise softwareWhat is Zimbra?
Zimbra Collaboration is an email and collaboration suite. Its internet-facing servers are frequent targets of exploitation.
Enterprise softwareWhat is FortiGate?
FortiGate is Fortinet's line of firewall and VPN appliances. Exploited edge devices are a common entry point and a place to hide C2.
Odin Research
Get the rules, not just the definition
Every Odin Research report ships with verified indicators and detection rules you can load today.