CVE-2026-93318
HighCVSS 7.5Exploitation Probability (EPSS)
Low risk5th percentile - higher than 5% of all known CVEs
Summary
A malicious image can advertise DiffIDs from another image while containing different layer contents. In affected versions, BuildKit could use the advertised DiffIDs to derive cache and snapshot identity without validating that they matched the actual layer contents. If a BuildKit daemon with shared or persistent cache first processes such a malicious image, a later build using the victim image may mount the attacker-controlled layer contents as the base image. This can allow code from the malicious image to run in the victim build, for example by replacing a commonly executed path such as /bin/sh. The attacker-controlled code may read build secrets mounted into the build, access other build resources, alter output artifacts, or hang the build. The issue affects both regular snapshotters and lazy-pulling snapshotters such as stargz.
Risk Assessment
An attacker can inject malicious code into the build process, potentially leading to secret theft, artifact integrity compromise, or disruption of CI/CD pipelines.
Recommendation
Upgrade BuildKit to a patched version and verify image integrity before using them in build processes.
Other vulnerabilities in BuildKit
See all- CVE-2026-93321Medium
A malicious frontend can submit an LLB definition that causes buildkitd to panic and terminate, interrupting all builds running on that daemon.
- CVE-2026-93315Medium
When proxy networking with CA injection is enabled, a build can modify its CA bundle before cleanup. This may cause cleanup to block, operate outside the build rootfs, or fail without failing the build.
- CVE-2026-97275Medium
Improper Validation of Specified Quantity in Input vulnerability in VillaTheme BuildKit – Product Builder for WooCommerce – Custom PC Builder woo-product-builder allows Input Data Manipulation.This issue affects BuildKit – Product Builder for WooCommerce – Custom PC Builder: from n/a through 1.0.28.
- CVE-2026-93323Medium
The Dockerfile frontend loaded Dockerfile and .dockerignore files into memory without a size limit. A build context with an oversized file could cause buildkitd to allocate memory proportional to the file size, potentially exhausting memory and terminating the daemon, interrupting other builds.
- CVE-2026-93322Medium
A malicious frontend can submit an LLB definition that causes buildkitd to panic and terminate, interrupting all builds running on that daemon.
- CVE-2026-93320Medium
BuildKit may be tricked into performing file actions with special file inodes where regular files are expected. Special files may block operations or, on rootful workers, allow unintended host device access.
- CVE-2026-93319Medium
A malicious external BuildKit frontend can send requests using the internal API that can create conditions for a data race that can cause the BuildKit daemon to panic.
- CVE-2026-93317Medium
An unauthenticated attacker controlling a registry or OCI-layout blob source could provide blob contents that did not match the claimed digest. The resulting snapshot could be cached under that digest and reused by a later victim build, compromising build-input integrity.
- CVE-2026-93316High
If BuildKit daemon is started with --cdi-disabled it can lead to daemon panic when builds try to use CDI devices. This can happen maliciously or by accident.
- CVE-2026-94298Medium
The BuildKit WordPress plugin before 1.0.29 does not properly sanitise and escape data submitted by contributor-level users before storing it and later using it in a SQL query, allowing a Contributor to inject SQL that runs against the database once the resulting content is published and viewed by any unauthenticated visitor.
Original NVD description (English source)
A malicious image can advertise DiffIDs from another image while containing different layer contents. In affected versions, BuildKit could use the advertised DiffIDs to derive cache and snapshot identity without validating that they matched the actual layer contents. If a BuildKit daemon with shared or persistent cache first processes such a malicious image, a later build using the victim image may mount the attacker-controlled layer contents as the base image. This can allow code from the malicious image to run in the victim build, for example by replacing a commonly executed path such as /bin/sh. The attacker-controlled code may read build secrets mounted into the build, access other build resources, alter output artifacts, or hang the build. The issue affects both regular snapshotters and lazy-pulling snapshotters such as stargz.
Vulnerability data from NVD (NIST) · CISA KEV · EPSS