Wiz Research uncovered a critical vulnerability (CVE-2026-3854) in GitHub's internal git infrastructure that could have affected both GitHub.com and GitHub Enterprise Server. By exploiting an injection flaw in GitHub's internal protocol, any authenticated user could execute arbitrary commands on GitHub's backend servers with a single git push command - using nothing but a standard git client.

Notably, this is one of the first critical vulnerabilities discovered in closed-source binaries using AI, highlighting a shift in how these flaws are identified. Despite the complexity of the underlying system, the vulnerability is remarkably easy to exploit. On GitHub.com, this vulnerability allowed remote code execution on shared storage nodes. We confirmed that millions of public and private repositories belonging to other users and organizations were accessible on the affected nodes. On GitHub Enterprise Server, the same vulnerability grants full server compromise, including access to all hosted repositories and internal secrets.

GitHub mitigated this issue on GitHub.com within 6 hours of our report, released patches for all supported versions of GitHub Enterprise Server, and published the CVE at the time of release. GitHub Enterprise Server customers should upgrade immediately - at the time of this writing, our data indicates that 88% of instances are still vulnerable. Detailed remediation steps and further technical details are available in GitHub’s security blog post.

GitHub greatly appreciates the collaboration, professionalism, and partnership that Wiz has shown throughout this process. A finding of this caliber and severity is rare, earning one of the highest rewards available in our Bug Bounty program, and serves as a reminder that the most impactful security research comes from skilled researchers who know how to ask the right questions. As the landscape evolves, these close partnerships with talented hunters and researchers are more important than ever.

This post breaks down the vulnerability, walks through the exploitation chain, and provides recommendations for GHES administrators to protect their environments.

Figure: Vulnerability overview - a single git push compromises GitHub's internal infrastructure

Figure: Vulnerability overview - a single git push compromises GitHub's internal infrastructure

Required Actions and Mitigations

GitHub.com: GitHub has mitigated this issue. No action is required for GitHub.com users.

GitHub Enterprise Server: Immediate action required.

Component Vulnerable Versions Fixed Version
GitHub Enterprise Server <= 3.19.1 3.14.24, 3.15.19, 3.16.15, 3.17.12, 3.18.6 and 3.19.3

Find Vulnerable GHES Instances with Wiz

Wiz customers can identify vulnerable GitHub Enterprise Server instances in their environments using this pre-built query in the Wiz Threat Center. The query identifies all GHES instances running a version vulnerable to this issue.

Figure: Wiz Threat Center query for vulnerable GHES instances

Figure: Wiz Threat Center query for vulnerable GHES instances

Why We Researched GitHub's Git Infrastructure

GitHub is the world's largest code hosting platform, home to hundreds of millions of repositories spanning open source projects, enterprise codebases, and critical infrastructure. Its internal git infrastructure-the pipeline that processes every git push-is one of the most security-sensitive systems on the internet. When a user pushes code, it passes through multiple internal services, each written in a different programming language. This multi-service architecture creates opportunities for inconsistencies in how each component parses and trusts shared data.

We've looked into GitHub Enterprise Server (GHES) in the past to hunt for these exact types of vulnerabilities. However, extracting and auditing the sheer volume of compiled blackbox binaries that run this pipeline historically required an impractical amount of time and manual effort.

But this is Round 2, and the landscape has shifted. By leveraging AI-augmented tooling-particularly automated reverse engineering using IDA MCP-we were able to do what was previously too costly. Using AI, we rapidly analyzed GitHub's compiled binaries, reconstructed internal protocols, and systematically identified where user input could influence server behavior across the entire pipeline. Thanks to this new capability, we found a fundamental flaw in how that input flows through GitHub's multi-service architecture.

Technical Deep-Dive