Is there a public PoC exploit available for EUVD-2026-53705?

Yes, a public proof-of-concept exploit is available for EUVD-2026-53705. Prioritise patching immediately. EPSS: 0.2%.

CowAgent EUVD-2026-53705

Q: What is the CVSS score of EUVD-2026-53705?

EUVD-2026-53705 has a CVSS 4.0 base score of 2.1 (Low). CVSS vector: CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:L/VI:L/VA:L/SC:N/SI:N/SA:N/E:P/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X. EPSS exploitation probability: 0.2%.

| CVE-2026-18992 LOW

Incorrect Authorization (CWE-863)

2026-08-06 VulDB

GHSA-wjcq-xp37-c947

Authentication Bypass Cowagent

2.1

CVSS 4.0 · Vendor: VulDB

Severity by source

Vendor (VulDB) PRIMARY

2.1 MEDIUM

CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:L/VI:L/VA:L/SC:N/SI:N/SA:N/E:P/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X

vuln.today AI

6.3 MEDIUM

Network-reachable with low-privilege authentication required; no scope change; all impacts low, consistent with contained authorization bypass within application boundary.

3.1 AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:L/A:L

4.0 AV:N/AC:L/AT:N/PR:L/UI:N/VC:L/VI:L/VA:L/SC:N/SI:N/SA:N

Primary rating from Vendor (VulDB).

CVSS VectorVendor: VulDB

Attack Vector

Network

Attack Complexity

Low

Privileges Required

Low

User Interaction

None

Scope

Lifecycle Timeline

Severity Changed

Aug 06, 2026 - 05:22 NVD

MEDIUM LOW

CVSS changed

Aug 06, 2026 - 05:22 NVD

5.3 (MEDIUM) 2.1 (LOW)

Analysis Generated

Aug 06, 2026 - 04:48 vuln.today

CVE Published

Aug 06, 2026 - 03:30 cve.org

MEDIUM 5.3

DescriptionCVE.org

A vulnerability was detected in zhayujie CowAgent up to 2.1.1. This vulnerability affects the function _select_tools of the file agent/evolution/executor.py of the component Self-Evolution Review Agent. Performing a manipulation results in incorrect authorization. The attack is possible to be carried out remotely. The exploit is now public and may be used.

AnalysisAI

Incorrect authorization in zhayujie CowAgent up to version 2.1.1 allows low-privilege authenticated remote attackers to bypass access controls in the Self-Evolution Review Agent component. The flaw resides in the _select_tools function of agent/evolution/executor.py (CWE-863), where authorization logic is implemented incorrectly, potentially permitting access to tools or agent capabilities outside the caller's intended permission scope. Publicly available exploit code exists (E:P in CVSS 4.0 vector), though active exploitation has not been confirmed by CISA KEV.

Technical ContextAI

CowAgent (cpe:2.3:a:zhayujie:cowagent:*:*:*:*:*:*:*:*) is a conversational AI agent framework developed by zhayujie that includes a Self-Evolution Review Agent - a subsystem responsible for dynamically selecting and orchestrating tools during agent execution. The vulnerable function _select_tools, located in agent/evolution/executor.py, implements CWE-863 (Incorrect Authorization): authorization logic is present but implemented incorrectly, meaning the system performs a check but reaches the wrong conclusion, as opposed to CWE-862 where no check exists at all. This distinction is important - the flaw is in the logic of the check, not its absence. The CVSS 4.0 vector (AV:N/AC:L/AT:N/PR:L) confirms network-reachable exploitation with low complexity and no special attack preconditions beyond holding a low-privilege account, consistent with an API or agent endpoint that trusts caller-supplied context insufficiently.

RemediationAI

No vendor-released patch has been identified at the time of analysis - no fixed version number is available from the provided data, advisory references, or the VulDB entry at https://vuldb.com/cve/CVE-2026-18992. The GitHub issue thread at https://github.com/zhayujie/CowAgent/issues/2904 and associated comment https://github.com/zhayujie/CowAgent/issues/2904#issuecomment-4756833239 should be monitored for upstream maintainer responses or merged fixes. As an immediate compensating control, operators should restrict access to the Self-Evolution Review Agent component by limiting which authenticated users can reach the agent/evolution executor endpoint - this reduces the PR:L attack surface to only trusted accounts. If the self-evolution feature is non-essential in your deployment, disabling it entirely via configuration eliminates the vulnerable code path without impacting core agent functionality. Audit all role assignments within CowAgent to enforce least-privilege access, ensuring low-privilege users cannot reach tool-selection endpoints. Watch the upstream repository for a tagged release addressing this issue before upgrading.