# SCA

**URL:** https://mazehq.com/solutions/sca
**Date:** 2026-11-19

![Maze dependency findings sorted by severity, from a Critical Apache Commons Text RCE down to Low.](https://mazehq.com/wp-content/uploads/2026/11/maze-sca-hero-findings-list-v2.png)

 

 

 

# Prioritize and fix exploitable dependencies

## Maze’s agents find every vulnerability in your dependencies, then investigate each one to prove what’s exploitable, not just reachable.

 [Book a demo](https://mazehq.com/demo)

 

 

 

 

---

Trusted by the **fastest growing** security teams

![Contentful](https://mazehq.com/wp-content/uploads/2026/06/contentful.svg)

![Zipline](https://mazehq.com/wp-content/uploads/2026/10/zipline.svg)

![Alloy](https://mazehq.com/wp-content/uploads/2026/06/alloy.svg)

![Halcyon](https://mazehq.com/wp-content/uploads/2026/06/halycon.svg)

![Cohere](https://mazehq.com/wp-content/uploads/2026/06/cohere-framed-1.svg)

![Forge](https://mazehq.com/wp-content/uploads/2026/06/forge-1.svg)

![PartsSource](https://mazehq.com/wp-content/uploads/2026/06/partssource-logo.svg)

 

 

 

---

![Maze marks an ejs template injection not exploitable, since the vulnerable option is never exposed to attacker input.](https://mazehq.com/wp-content/uploads/2026/11/maze-sca-reachable-not-exploitable-v2.png)

 

 

 

### Exploitable, not reachable

Maze agents go beyond reachability to prove what’s exploitable, checking whether an attacker controls what reaches the vulnerable code in your application.

 

 

 

 

---

![Maze traces a lodash prototype pollution from an SQS consumer message body into lodash.merge, marked exploitable.](https://mazehq.com/wp-content/uploads/2026/11/maze-sca-call-path-v2.webp)

 

 

 

### Beyond traditional call graphs

Maze’s agents trace every path through your code and every open-source package in between, in any language you write. Traditional reachability results are often incomplete.

 

 

 

 

---

![Maze rates a PyYAML RCE exploitable but low likelihood, because its route sits inside a private EKS cluster.](https://mazehq.com/wp-content/uploads/2026/11/maze-sca-severity-likelihood.png)

 

 

 

### Risk based on your context

Once agents prove a finding is exploitable, they weigh its impact and how realistic the attack is, using the full context of your code, cloud, and runtime. A package running in production and one on a staging service shouldn’t be treated the same.

 

 

 

 

---

![Maze dependency findings in the Investigating state, with no severity assigned yet while Maze works through them.](https://mazehq.com/wp-content/uploads/2026/11/maze-sca-findings-investigating.webp)

 

 

 

### Reasoning without frontier cost

Maze’s agents continually investigate your dependencies, bringing frontier-level reasoning at a cost you can afford, without credit or token limits.

 

 

 

 

---

### Agents investigate every dependency

    Every package, traced to the code that calls it  Agents find the manifest that pulled in each package, the exact version your build used, and the function that calls it, for direct and transitive dependencies alike.![Maze traces a golang-jwt vulnerability from the vulnerable dependency to a reachable refresh route.](https://mazehq.com/wp-content/uploads/2026/11/maze-sca-traced-to-route.png)

 

    Agents that research every CVE  Agents research the fix behind every CVE and check the package your build pulled, instead of the version in your manifest, so you know whether the vulnerable code is gone.![Maze shows a follow-redirects vulnerability as not exploitable, because the build installs 1.15.9 with the fix.](https://mazehq.com/wp-content/uploads/2026/11/maze-sca-fix-landed.png)

 

    Every pull request investigated with a team of agents  Maze looks for vulnerabilities on every pull request and investigates them with the full context of your code and cloud, so it only blocks merges on real risk and false positives don’t slow you down.![Maze blocks a GitHub pull request, with a comment marking node-forge exploitable and minimist not exploitable.](https://mazehq.com/wp-content/uploads/2026/11/maze-sca-pull-request.png)

 

 

 

 

![Maze traces a golang-jwt vulnerability from the vulnerable dependency to a reachable refresh route.](https://mazehq.com/wp-content/uploads/2026/11/maze-sca-traced-to-route.png)

 

 

![Maze shows a follow-redirects vulnerability as not exploitable, because the build installs 1.15.9 with the fix.](https://mazehq.com/wp-content/uploads/2026/11/maze-sca-fix-landed.png)

 

 

![Maze blocks a GitHub pull request, with a comment marking node-forge exploitable and minimist not exploitable.](https://mazehq.com/wp-content/uploads/2026/11/maze-sca-pull-request.png)

 

 

 

 

---

  

###### Find and fix
exploitable dependencies

 [Book a demo](https://mazehq.com/demo)

 

---

## Frequently Asked Questions

Talk to us about how Maze’s agents
investigate your dependencies.

 [Get in touch](https://mazehq.com/contact-us)

 

 

#### What is Maze AI SCA?![](https://mazehq.com/wp-content/themes/okd/static/img/accordion_plus.svg)

Maze AI SCA finds every vulnerability in your open-source dependencies, direct and transitive, and investigates each one the way your best security engineer would. Agents trace each package to the code that calls it and prove whether it’s exploitable in your environment, and the ones that are come with evidence and a fix.

#### How is this different from the SCA tool I already run?![](https://mazehq.com/wp-content/themes/okd/static/img/accordion_plus.svg)

Most SCA tools use rules to find vulnerabilities and then rank them by CVE score. Some add reachability on top. Maze goes a step further, with agents checking whether an attacker can control what reaches the vulnerable code in your application, so you only fix what’s exploitable.

#### Isn’t reachability enough?![](https://mazehq.com/wp-content/themes/okd/static/img/accordion_plus.svg)

Reachability tells you the vulnerable code can be called, not whether an attacker can trigger the flaw. In one investigation, a vulnerable library was reachable five different ways, but every request it made went to a fixed internal service, so an attacker had no way to send it a malicious payload. Instead of another urgent ticket, that finding came off the list with the evidence to show why.

#### Is Maze Code a scanner?![](https://mazehq.com/wp-content/themes/okd/static/img/accordion_plus.svg)

Yes. Maze Code scans for vulnerable dependencies (SCA), flaws in your own code (SAST), secrets, licenses, and more, so you don’t need a separate scanner to get started.

#### Does it cover transitive dependencies?![](https://mazehq.com/wp-content/themes/okd/static/img/accordion_plus.svg)

Yes. Agents trace the full call path through your code and every open-source package in between.

#### What languages do you support?![](https://mazehq.com/wp-content/themes/okd/static/img/accordion_plus.svg)

Because Maze uses AI to find and investigate vulnerabilities instead of static rules, it isn’t limited to the languages a rule set was written for. That includes older languages like C++.

#### Does it run in my pull requests?![](https://mazehq.com/wp-content/themes/okd/static/img/accordion_plus.svg)

Yes. Maze investigates the packages every pull request adds and blocks merges on real risk. Unlike other solutions, pull requests get the same agents and the same investigation, without adding time. Fixes route to the developer who owns the code or to the coding agents your team already uses.

#### How does pricing work?![](https://mazehq.com/wp-content/themes/okd/static/img/accordion_plus.svg)

We aim to keep pricing simple and fair and offer an annual contract. Large language models aren’t cheap, but we’ve worked hard to optimize cost and performance to make sure our pricing is reasonable.

#### Is Maze hosted in the cloud?![](https://mazehq.com/wp-content/themes/okd/static/img/accordion_plus.svg)

Yes, Maze is cloud-hosted in AWS. Customers can choose between multi-tenant and single-tenant hosting.

#### What compliance standards do you meet?![](https://mazehq.com/wp-content/themes/okd/static/img/accordion_plus.svg)

Maze has been built for enterprise from day one. We’re ISO 27001 and SOC 2 Type II certified.

 

 

 

---

---

## Autonomous security
from commit to production

 

---

  [ ### Dependencies (AI SCA)

Find, triage, and remediate CVEs in third-party code, using your existing scanner or ours.

 

  ](https://mazehq.com/platform/code) [ ### Your Code (AI SAST)

Find and fix business-logic vulnerabilities that SAST misses, or triage results from your existing scanner.

 

  ](https://mazehq.com/platform/code) [ ### Cloud Vulnerabilities

Triage and remediate CVEs in containers and VMs, and catch zero-days before scanners see them.

 

  ](https://mazehq.com/platform/cloud) [ ### Why Maze?

Everyone uses AI now. What makes Maze different, and how is the platform built for AI agents?

 

  ](https://mazehq.com/platform) [ ### Our Story

Why we’re building Maze to be a security company that thinks differently.

 

  ](https://mazehq.com/our-story) [ ### Resources

Security research, blogs, video, and more from security leaders and the Maze team.

 

  ](https://mazehq.com/resources) 

 

---

   Resources 

Selected Resources

 

[View all resources](https://mazehq.com/resources)

 

 

 

---

  [[Vulnerabilities]

## Java4Shell: how to check exploitability on every JVM you run

CVE-2026-114938, known as “Java4Shell”, is a heap out-of-bounds write in OpenJDK’s AVX-512 Base64 decoder. Calif, who found it, reports chaining it to remote code execution in default installs…

 November 18, 2026  

 

 ![White Java logo in a dark browser window](https://mazehq.com/wp-content/uploads/2026/11/java4shell-java-logo-hero-1024x764.png) 

 ](https://mazehq.com/blog/java4shell) 

 

- [[Vulnerabilities]
    
    ### PHP2Shell: two critical PHP RCEs, and how to tell if you’re affected
    
     
    
     October 22, 2026  
    
     ](https://mazehq.com/blog/php2shell)
- [[Vulnerabilities]
    
    ### Removed from a Plane project? One invite can bring you back as admin
    
     
    
     October 20, 2026  
    
     ](https://mazehq.com/blog/plane-access-control-bug)
- [[Company]
    
    ### How Cohere Health cleared 92% of its critical and high findings and got their time back
    
     
    
      
    
     ](https://mazehq.com/customer-story/how-cohere-health-cleared-92-of-its-critical-and-high-findings)

 

 

---