Two serious vulnerabilities in the n8n automation platform could allow attackers to compromise systems, access sensitive information, and execute arbitrary code. Identified as CVE-2026-1470 and CVE-2026-0863, these flaws were discovered by researchers at JFrog. CVE-2026-1470 has a critical severity score of 9.9 out of 10 due to a sandbox escape that lets malicious code run on the main n8n node. Although it requires authentication, the flaw can still be exploited by non-admin users who have permissions to modify workflows, allowing them to gain control over the underlying infrastructure. CVE-2026-0863 involves a similar sandbox escape in Python, which also permits execution of OS commands on the main node. Both vulnerabilities highlight the challenges of securely sandboxing high-level languages like JavaScript and Python, even with multiple security layers in place. Users running self-hosted versions of n8n are particularly at risk, while the cloud version has been patched. The vulnerabilities were resolved in various n8n versions, and users are urged to upgrade promptly. Recent scrutiny of n8n follows the disclosure of another severe flaw, "Ni8mare," which allows unauthenticated access to local instances. Despite ongoing threats, many users seem slow to patch their systems. Scans indicated that about 39,900 instances remain vulnerable, down from 60,000, suggesting that remediation efforts are not keeping pace with the risks. Security researchers are concerned that the public availability of proof-of-concept exploits could lead to increased targeting of unpatched n8n deployments.