[](https://devin.ai/blog)

# Checking for NGINX2Root and Java4Shell with Threats in Devin Security

CognitionNovember 18, 20263 min read

AI is rapidly increasing the volume of new vulnerabilities discovered. Just this week, we saw two new sets of vulnerabilities in foundational software disclosed:

1. **[Java4Shell](https://calif.io/java4shell)**, published November 17, is a vulnerability in the Java runtime that can lead to memory disclosure and remote code execution.
2. **[NGINX2Root](https://xint.io/blog/nginx2root-remote-code-execution)**, published November 18, is a set of five memory-corruption bugs in nginx that can also lead to remote code execution.

To make it easier for enterprises to respond to this growing volume, we’re launching Devin Security Threats: a managed index of new vulnerabilities disclosed, integrated with Devin for one-click scanning and remediation.

## How threats work

A threat is a description of something that could go wrong in your systems: what is affected, how an attacker would use it, and what fixes it. Cognition publishes threats for major disclosures like these two, and your team can add its own.

![Devin Security Overview showing NGINX2Root and Java4Shell as threats published by Cognition, each with View threat and Scan with Devin buttons](https://devin.ai/images/blog/nginx2root-java4shell-threats/overview-threat-cards.png)

**Scan with Devin** opens a new scan with the threat already selected under **Focus on threats**. You can select the repositories that you would like Devin to scan. The repositories are scanned with the same [Agentic MapReduce](https://devin.ai/blog/agentic-map-reduce) process as any Devin Security scan.

Findings are then ranked against your threat model. Once the findings are found, you can assign the findings to Devin and it opens pull requests for your team to review. With **Auto-triage** on, you can pick the repositories and a minimum severity, and Devin scans them for each new threat as it is published. Optionally, a threat whose scan finds nothing is dismissed automatically.

Threats are also not limited to what Cognition publishes. If a bug bounty report, a vendor notice or an incident review raises a question about your own code, create a threat with a title and a description of what could happen and how, and scan for it the same way.

## Getting started

Both threats are live now at [app.devin.ai/security](https://app.devin.ai/security).

[Learn more about Devin Security](https://devin.ai/security)

## Fixing with skills

If you would rather run the remediation as a Devin session than as a scan, our [Vulnerability Fixes plugin](https://devin.ai/security/vulnerability-fixes) now has a skill for each of these:

- [`fix-nginx-vuln`](https://github.com/CognitionAI/devin-marketplace/tree/main/plugins/devin-vuln-fixes/skills/fix-nginx-vuln/SKILL.md) for the five NGINX2Root bugs.
- [`fix-java-vuln`](https://github.com/CognitionAI/devin-marketplace/tree/main/plugins/devin-vuln-fixes/skills/fix-java-vuln/SKILL.md) for Java4Shell.

Install the plugin from the Devin Marketplace, or from the Devin CLI:

```
devin plugins install CognitionAI/devin-marketplace#plugins/devin-vuln-fixes
```

Then start a session:

[Fix nginx vulnerabilities](https://app.devin.ai/?prompt=Use+the+fix-nginx-vuln+skill+from+the+Vulnerability+Fixes+plugin.+If+the+plugin+is+not+installed+already%2C+add+it.+Then+inventory+every+repo+connected+to+Devin+for+nginx+deployments+affected+by+NGINX2Root+%28CVE-2026-123007%2C+CVE-2026-123633%2C+CVE-2026-121810%2C+CVE-2026-123634%2C+CVE-2026-113425%29%2C+show+me+the+table%2C+and+wait+for+my+approval+before+patching.)

[Fix Java vulnerabilities](https://app.devin.ai/?prompt=Use+the+fix-java-vuln+skill+from+the+Vulnerability+Fixes+plugin.+If+the+plugin+is+not+installed+already%2C+add+it.+Then+inventory+every+repo+connected+to+Devin+for+Java+runtimes+affected+by+Java4Shell+%28CVE-2026-114938%29%2C+show+me+the+table%2C+and+wait+for+my+approval+before+patching.)

The skills are plain Markdown and open source, so you can also read them as a checklist or give them to another coding agent.

Table of contents

[How threats work](#how-threats-work)[Getting started](#getting-started)[Fixing with skills](#fixing-with-skills)
