Wazuh Manager CVE-2026-74034

Requires the attacker to be an enrolled Wazuh agent (possessing valid enrollment/client.keys) able to send messages over the agent-to-manager channel; the attacker then issues an upgrade_custom command with an attacker-controlled file_path, which the manager's wm_agent_upgrade processes without an origin check and reads as root. Affects wazuh-manager 4.1.0 through 4.14.7 in default configuration (no special feature toggle is indicated). LIMITING FACTORS: despite the vendor's PR:N rating, exploitation realistically depends on obtaining agent enrollment (a compromised/rogue agent or stolen keys), so it is not anonymous internet-wide; no user interaction is needed, and the fix in 4.14.8 restricts which upgrade commands the agent channel will accept.

Signals point to a real, high-severity confidentiality issue but one whose practical exploitation is gated. The vendor CVSS 4.0 base score is 8.7 (High) with vector AV:N/AC:L/PR:N/UI:N and VC:H (high confidentiality impact, no integrity or availability impact), reflecting arbitrary file read as root. The flaw was reported by VulnCheck (a credible source) and a vendor patch exists (v4.14.8), which raises confidence the vulnerability is genuine. However, no EPSS score, CISA KEV listing, or public POC is present in the provided data, so there is no evidence of active exploitation or automated weaponization - this is a priority to patch but not (on available signals) an under-active-attack emergency. CONFLICT TO VERIFY: the CVSS vector asserts PR:N (unauthenticated), yet the description states the attacker must be an 'enrolled agent' issuing messages over the agent channel, which in practice requires valid agent enrollment/client.keys - this discontinuity means realistic exploitation presupposes a compromised or rogue registered agent rather than an anonymous internet attacker, lowering the real-world likelihood of mass exploitation relative to the raw score. The confidentiality impact is nonetheless serious because exfiltrating client.keys enables agent impersonation and lateral movement across the fleet, and /etc/shadow exposure aids credential cracking.

An attacker who controls an enrolled Wazuh agent (e.g., a compromised endpoint or stolen agent keys) sends a crafted upgrade_custom message over the agent-to-manager channel with an attacker-controlled file_path. Because wm_agent_upgrade lacks an origin check, the manager processes the command and reads the specified file as root, returning sensitive contents such as /etc/shadow or client.keys. No public exploit was identified at time of analysis, but the attack is low-complexity once agent enrollment is obtained.