GHSA-2J9V-P4XJ-CJW2
Vulnerability from github – Published: 2026-08-14 19:24 – Updated: 2026-08-14 19:24Impact
On an instance of Lima running with qemu driver, an arbitrary user in the VM could access /run/lima-guestagent.sock when the guest agent is enabled.
This could result in running an arbitrary command with the root privileges in the VM (not on the host), as lima-guestagent.sock provides the tunneling service for an arbitrary address, including a Unix socket address for privileged daemons like D-Bus.
This vulnerability is not exploitable on vz driver, as the guest agent uses vsocks instead of Unix sockets.
Patches
Patched in Lima v2.1.3 (8a45892378d22f40505c31a38f786a07701b6d50)
[!NOTE] The default user account in the VM can still run an arbitrary command as the root via the guest agent socket. This is not a vulnerability, as the user can already run an arbitrary command with
sudoby design.
Workarounds
- On macOS hosts, use
vzdriver instead ofqemu(limactl create --vm-type=vz. Default since v1.0.) - Or, disable the guest agent (
limactl create --plain)
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 2.1.2"
},
"package": {
"ecosystem": "Go",
"name": "github.com/lima-vm/lima/v2"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "2.1.3"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-53657"
],
"database_specific": {
"cwe_ids": [
"CWE-276",
"CWE-668"
],
"github_reviewed": true,
"github_reviewed_at": "2026-08-14T19:24:33Z",
"nvd_published_at": "2026-07-10T17:16:57Z",
"severity": "HIGH"
},
"details": "### Impact\nOn an instance of Lima running with `qemu` driver, an arbitrary user in the VM could access `/run/lima-guestagent.sock` when the guest agent is enabled.\n\nThis could result in running an arbitrary command with the root privileges in the VM (**not on the host**), as `lima-guestagent.sock` provides the tunneling service for an arbitrary address, including a Unix socket address for privileged daemons like D-Bus.\n\nThis vulnerability is not exploitable on `vz` driver, as the guest agent uses vsocks instead of Unix sockets.\n\n### Patches\nPatched in Lima v2.1.3 (8a45892378d22f40505c31a38f786a07701b6d50)\n\n\u003e [!NOTE]\n\u003e The default user account in the VM can still run an arbitrary command as the root via the guest agent socket.\n\u003e This is not a vulnerability, as the user can already run an arbitrary command with `sudo` by design.\n\n### Workarounds\n- On macOS hosts, use `vz` driver instead of `qemu` (`limactl create --vm-type=vz`. Default since v1.0.)\n- Or, disable the guest agent (`limactl create --plain`)",
"id": "GHSA-2j9v-p4xj-cjw2",
"modified": "2026-08-14T19:24:33Z",
"published": "2026-08-14T19:24:33Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/lima-vm/lima/security/advisories/GHSA-2j9v-p4xj-cjw2"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-53657"
},
{
"type": "WEB",
"url": "https://github.com/lima-vm/lima/commit/8a45892378d22f40505c31a38f786a07701b6d50"
},
{
"type": "WEB",
"url": "https://github.com/lima-vm/lima/commit/b08cae8a670cf916d5da11c48a6de76dabd89678"
},
{
"type": "PACKAGE",
"url": "https://github.com/lima-vm/lima"
},
{
"type": "WEB",
"url": "https://github.com/lima-vm/lima/releases/tag/v2.1.3"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:L/AC:L/PR:H/UI:N/S:C/C:H/I:H/A:H",
"type": "CVSS_V3"
}
],
"summary": "Lima: An arbitrary user in a QEMU VM could gain the root privilege in the VM via the guest agent socket"
}
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.