Security researchers have disclosed two critical vulnerabilities in the open-source OpenCode developer tool that enable malicious websites to hijack a user’s machine simply by visiting a crafted webpage. These flaws allow attackers to execute arbitrary code and interact with local system files through OpenCode’s exposed APIs, leading to remote code execution (RCE) with the privileges of the logged-in user. The disclosed vulnerabilities in OpenCode arise from insecure architectural decisions in its local web interface and API server, which are enabled by default when the tool is running. OpenCode exposes a locally bound HTTP service intended for developer interaction; however, this service lacks proper authentication controls and input validation. As a result, untrusted web content can interact directly with sensitive local endpoints without user awareness, creating a severe security boundary violation between the browser and the host operating system. One of the flaws, CVE-2026-22813, is caused by improper sanitization of HTML content rendered within the OpenCode web UI. This allows attackers to inject malicious JavaScript through crafted chat messages or content delivered from attacker-controlled sources. Once executed in the OpenCode interface, the script can issue authenticated requests to the local API server, effectively granting the attacker the ability to execute shell commands, manipulate sessions, and access system resources with the same privileges as the user. The second vulnerability, CVE-2026-22812, stems from the absence of authentication and overly permissive Cross-Origin Resource Sharing (CORS) policies on the OpenCode HTTP server. The server accepts requests from any origin and exposes powerful endpoints that allow command execution and file access. This enables drive-by exploitation, where simply visiting a malicious website while OpenCode is running can lead to full system compromise. Together, these flaws allow remote attackers to hijack developer machines without requiring user interaction beyond normal web browsing.
Android users are being targeted in a large-scale malware operation designed to silently enroll victims into premium mobile billing services without consent. The campaign abuses ca...
U.S. and Canadian law enforcement agencies have detained an alleged operator connected to the KimWolf botnet, a cybercrime platform known for conducting large-scale DDoS attacks us...
A newly identified Phishing-as-a-Service (PhaaS) platform named Kali365 is actively targeting Microsoft 365 users by abusing OAuth device authentication flows to bypass multi-facto...