Executive brief
A security vulnerability exists in idachev mcp-javadc, a tool used for Java decompilation via the Model Context Protocol (MCP). An attacker can remotely execute unauthorized commands on the server hosting this service by sending a specially crafted request. This could lead to a complete system takeover, allowing unauthorized access to sensitive data or disruption of operations.
Technical details
An OS command injection vulnerability (CWE-78) exists in idachev mcp-javadc versions up to 1.2.4 within the index.js component. The vulnerability is located in the decompile-from-jar tool, where the user-supplied jarFilePath argument is insufficiently sanitized before being interpolated into shell commands using execPromise (e.g., jar tf and jar xf). A remote attacker can break out of the double-quoted string by injecting shell metacharacters, leading to arbitrary command execution with the privileges of the server process. While a public exploit exists, the maintainer has not yet released a patch.
Affected products
- idachev mcp-javadc up to 1.2.4
Timeline
- 2026-03-15: disclosed: Vulnerability reported to the vendor via issue report.
- 2026-03-24: other: Public exploit published on GitHub.
- 2026-04-08: advisory: CVE-2026-5802 published.