JumpServer CVE-2026-112159

Requires an authenticated JumpServer account (PR:L) with permission to create or edit a Host asset, supplied value being a crafted IP/HOST address containing Jinja2 template syntax ({{ }}, {% %}, or {

#}); the payload executes only when that address is subsequently processed by the Template Rendering component (the UI:P passive-interaction condition). Limiting factors: affects versions up to and including 4.10.19 (fixed in 5.0.0); exploitation is not possible against anonymous/unauthenticated users; and it depends on the asset address reaching the vulnerable rendering path. These constraints mean exploitation is realistically available to insiders or attackers who have already obtained valid low-privileged credentials.

This is a moderate, authentication-gated risk rather than a drop-everything emergency. The vendor-supplied CVSS 4.0 vector (AV:N/AC:L/AT:N/PR:L/UI:P) indicates a remote, low-complexity attack that nonetheless requires an authenticated account (PR:L) and some passive user interaction (UI:P) - meaning an attacker must already be a valid JumpServer user with the ability to create or edit an asset, and the malicious IP/HOST value must subsequently be rendered. A public proof-of-concept exists (VulDB gist), which raises practical exploitability, but the CVE is NOT in CISA KEV and no EPSS score was provided, so there is no evidence of active, widespread exploitation - exploitation status is 'publicly available exploit code exists,' which is distinct from confirmed in-the-wild use. The most significant signal conflict is impact magnitude: the vendor rates confidentiality/integrity/availability as Low (VC:L/VI:L/VA:L, base score 5.1), yet the root cause is CWE-78 OS command injection via server-side template (Jinja) rendering, confirmed by the patch adding validators that strip Jinja delimiters from the address field - OS command injection normally yields high-impact code execution. This divergence should be verified with the vendor: if the injected command runs with meaningful privileges, real-world severity is materially higher than the 5.1 Medium rating suggests.

An authenticated JumpServer user with asset-management permissions creates or edits a Host asset and places a malicious Jinja2 template payload (e.g. {{ ... }} or {% ... %}) in the IP/HOST address field. When the platform later performs template rendering on that address, the embedded expression is evaluated server-side and injected OS commands execute on the JumpServer host. A public proof-of-concept exploit is available, lowering the skill barrier for abuse by any low-privileged account.