Debuggix Ran 9 Security Engines Across 100 Repos. Here Is The Raw Data On Dependency CVEs.

3 7 22
calendar_today agoschedule3 min read

The Debuggix team ran a security experiment across 100 public GitHub repositories. We used 9 engines running in parallel: Semgrep, Bandit, Gitleaks, TruffleHog, Trivy, ESLint, Hadolint, Checkov, and OSV-Scanner.

The goal was to collect raw data on the actual state of dependency security. Not marketing claims. Not vendor reports. Just findings from real codebases.

Here is what the data showed.


Finding One: Dependency CVEs Are Universal

Every single repository scanned had at least one dependency CVE.

Not most. Not 99 percent. All 100 repositories.

This included projects with hundreds of thousands of stars. Projects maintained by full-time security teams. Projects that had been scanned before. Projects built entirely by AI three weeks ago.

The most common vulnerable packages were protobufjs, xmldom, axios, and Hono.

Protobufjs has over 4 million weekly downloads on npm. The CVEs found relate to prototype pollution and denial of service. Prototype pollution can lead to remote code execution in some contexts.

Xmldom has been deprecated but remains in thousands of projects. The CVEs found include XML external entity injection, which can lead to local file disclosure on the server.

Axios has over 20 million weekly downloads. The CVEs found include server-side request forgery and insecure redirect handling. An attacker can use SSRF to access internal services not exposed to the internet.

Hono has grown rapidly among edge compute developers. The CVEs found include request smuggling and improper input validation.


Finding Two: The Severity Distribution

Across all 100 repositories, the Debuggix scan found:

  • Critical severity CVEs: 12 percent of findings
  • High severity CVEs: 28 percent of findings
  • Medium severity CVEs: 45 percent of findings
  • Low severity CVEs: 15 percent of findings

Critical and high severity findings appeared in 73 percent of repositories. These are vulnerabilities that can lead to remote code execution, data breach, or complete system compromise.

The remaining 27 percent of repositories had only medium and low severity findings. These are still vulnerabilities. They still need fixing. But they are less likely to be exploited immediately.


Finding Three: Maintainers Did Not Know

When Debuggix approached maintainers with a list of specific CVEs affecting their projects, the response was consistent.

Almost every maintainer said some version of "I did not know that package was vulnerable."

This is not a failure of individual developers. It is a failure of the ecosystem. Developers cannot fix what they do not know exists. And most developers have never run a dependency scan.

The difference between a secure project and an insecure project is not code quality. It is awareness.


What You Can Run Right Now

You do not need a paid tool to check your dependencies. These commands are free and run locally.

For npm:

npm audit

For Yarn:

yarn audit

For Python:

pip-audit

For Rust:

cargo audit

For Go:

govulncheck

Each command takes less than 30 seconds. Run one on your project today.


How Debuggix Scans Dependencies

Debuggix runs three dependency scanning engines in parallel: Trivy, OSV-Scanner, and an integration with Snyk's open source vulnerability database.

Trivy excels at container scanning but also covers language-specific dependencies. OSV-Scanner uses Google's open source vulnerability database, which includes CVEs from the GitHub Advisory Database and RustSec. The Snyk integration adds coverage from their commercial database.

Running multiple engines catches what single engines miss. One database might have a CVE that another does not. One engine might flag a vulnerability that another classifies as low priority.

The result is a comprehensive dependency report showing every known CVE affecting your project, with severity ratings and suggested version upgrades.

Debuggix is free for open source repositories. Paid plans for private repos start at $29 per month.

Try it: debuggix.space

🔥 Join developers growing publicly
Share your knowledge, build in public, and grow your developer presence with a global community.

More Posts

Debuggix Analyzed AI-Generated Code From Cursor, Lovable, And Bolt. Here Are The 5 Security Patterns

Lucky - Jun 12

Your Backup Data Knows More Than You Think. HYCU aiR Is Finally Asking It the Right Questions.

Tom Smithverified - May 14

The Sovereign Vault — A Comprehensive Guide to Protocol-Driven AI

Ken W. Algerverified - Jun 4

Debuggix Tested 9 Security Engines On Kubernetes Goat. 134 Raw Findings. Only 6 Were Real. Here Is

Lucky - Jun 12

The 2026 State of GitHub Security: What 100 Repos Taught Me About Dependency CVEs and AI Code

Lucky - Jun 11
chevron_left
520 Points32 Badges
9Posts
7Comments
1Connections

Related Jobs

View all jobs →

Commenters (This Week)

8 comments
7 comments
2 comments

Contribute meaningful comments to climb the leaderboard and earn badges!