---
id: "CVE-2026-116964"
kind: "vulnerability"
aliases: ["CVE-2026-116964"]
title: "konflux-ci: repository writers can run code in CI/CD pipelines"
severity: "high"
cvss: 8.4
epss: null
exploited: false
status: "open"
category: "other"
products: []
ecosystems: []
tags: ["ci-cd", "code-injection", "github-actions", "konflux-ci", "kubernetes", "other"]
not_affected_if: ["manual-pr-review", "no-konflux-ci", "no-repo-write-access"]
published_at: "2026-10-31T09:16:33+00:00"
first_seen_at: "2026-10-31T09:34:41+00:00"
updated_at: "2026-10-31T09:37:09+00:00"
sources: [{"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-116964", "first_seen_at": "2026-10-31T09:34:41+00:00"}, {"source": "analyst", "url": null, "first_seen_at": "2026-10-31T09:37:09+00:00"}]
---

# konflux-ci: repository writers can run code in CI/CD pipelines

konflux-ci has a flaw where shell commands injected into environment export scripts are evaluated unchecked by manifest synchronization scripts. An attacker with repository write access can run arbitrary code in the CI/CD workflow and reach stored tokens and deployment credentials.

_Summary, check and fix written by a model from the sources; labelled as such._

## Timeline

- 2026-10-31T09:34:41+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-116964
- 2026-10-31T09:37:09+00:00 analyst — Red Hat CVE page read (CWE-94, eval() in manifest scripts, CVSS 8.4). It gives no affected versions or fixed release, and the konflux-ci project has no advisory I could find. Red Hat says the only mitigation is human review of PRs. Fix stays unknown; nothing verified. To settle it, check konflux-ci GitHub commits or releases for the eval removal, or Bugzilla 2564209.

## Documents

- [advisory] CVE-2026-116964 - Red Hat Customer Portal https://access.redhat.com/security/cve/cve-2026-116964

## From the sources

A flaw was found in konflux-ci. An attacker with repository write access can achieve arbitrary code execution within the Continuous Integration and Continuous Delivery (CI/CD) workflow environment. By injecting shell commands into environment export scripts, the attacker causes manifest synchronization scripts to dynamically evaluate unvalidated script output. Successful exploitation allows the attacker to compromise CI/CD pipelines, steal sensitive authentication tokens and deployment credentials, and gain unauthorized access to production systems.

---

_Free to read at a human pace. An API with keys and an MCP server are in preparation; early access: hello@infrastack.ai_