Alex Joseph
A Critical Flaw in n8n’s AI Agents Feature
A significant security flaw has been identified in n8n, a popular workflow automation tool. This vulnerability, tracked as CVE-2026-65015, allows a user with limited “Project Viewer” access to gain higher privileges within the system. This means an unauthorized user could potentially run harmful commands or access sensitive information that should be off-limits to them.
CVE Details
- Product: n8n
- Published Date: July 22, 2026
- Severity: High
- Status: Analyzed
Affected Products
The privilege escalation vulnerability impacts n8n versions across both its community and enterprise editions. Specifically, any n8n version prior to 2.29.8 and 2.30.1 is vulnerable. This includes the n8n community and enterprise editions built on Node.js.
Current Status
This vulnerability has been officially analyzed. This means the details are confirmed and understood by the cybersecurity community.
Severity Level
Rated as “High” severity with a CVSS score of 8.8, this vulnerability poses a serious risk. A high severity indicates that exploiting this flaw could lead to a significant impact, such as unauthorized access to confidential data or the ability to execute malicious code on the server. The ease of exploitation, requiring only low privileges and no user interaction (beyond chatting with an AI agent), contributes to its critical rating.
Possible Solutions
To protect your n8n installations, it’s crucial to take immediate action:
Patches
The most effective solution is to update your n8n instance to a patched version. The issue has been resolved in versions n8n 2.29.8 and n8n 2.30.1, or any later releases. Updating ensures that the underlying flaw in the AI Agents feature is corrected.
Workarounds (Temporary Mitigations)
If an immediate upgrade is not feasible, consider these temporary steps:
- Disable the AI Agents module: Remove
agentsfrom yourN8N_ENABLED_MODULESenvironment variable to completely turn off the problematic feature. - Restrict project access: Limit “Project Viewer” roles strictly to fully trusted individuals. Avoid giving this access to any untrusted users on projects that utilize AI agents with node tools enabled.
- Disable dangerous nodes: If enabled, deactivate command-execution nodes (like “Execute Command” or “SSH”) to reduce the potential impact of an exploit.
Please remember, these workarounds offer only partial protection and should only be used as a short-term measure until you can apply the official patches.
References
https://github.com/n8n-io/n8n/security/advisories/GHSA-x5vx-c2c8-m3w9
https://www.vulncheck.com/advisories/n8n-before-privilege-escalation-via-run-node-tool
