Kubeflow Pipelines 允许用户构建和部署可移植、可扩展的机器学习工作流。在 2.18.0 版本之前,当数据源为 或 时,前端工件下载路径会接受 参数;同时,归档的 Pod 日志路径也会接受来自工作流状态的 S3 兼容端点元数据,而无需这些端点与由操作员控制的存储配置相匹配。在随附的已禁用授权的独立部署配置中,未经验证的调用者可以诱导共享的 Pod 连接到任意可访问的主机并返回响应数据;此外,若请求使用已配置的对象存储凭据进行签名,还可能泄露访问密钥标识符。基于多用户密钥的元数据仍受限于服务器命名空间
Although we use advanced large model technology, its output may still contain inaccurate or outdated information.Shenlong tries to ensure data accuracy, but please verify and judge based on the actual situation.
| # | POC Description | Source Link | Shenlong Link |
|---|
No public POC found.
Login to generate AI POC| CVE-2026-77596 | 9.6 CRITICAL | Kubeflow Pipelines API server lacks authorization for shared pipeline uploads |
| CVE-2026-88980 | 8.6 HIGH | Kubeflow Pipelines: Unauthenticated server-side request forgery in the frontend http/https |
| CVE-2026-88984 | 7.2 HIGH | Kubeflow Pipelines: RBAC bypass on shared pipelines allows any authenticated user to delet |
| CVE-2026-88985 | 6.5 MEDIUM | Kubeflow Pipelines: SQL injection through unvalidated pagination tokens in Kubeflow Pipeli |
| CVE-2026-88986 | 6.1 MEDIUM | Kubeflow Pipelines: Reflected XSS in the frontend pod information and events error respons |
| CVE-2026-88979 | 5.8 MEDIUM | Kubeflow Pipelines: Unauthenticated SSRF vulnerability in Kubeflow Pipelines TensorBoard p |
| CVE-2026-88982 | 5.4 MEDIUM | Kubeflow Pipelines: Cross-tenant execution-cache poisoning / output substitution via a nam |
| CVE-2026-88983 | 5.0 MEDIUM | Kubeflow Pipelines — API server fetches a caller-supplied pipelineURL without host validat |