Roo Code is an AI-powered autonomous coding agent. The project-specific MCP configuration for the Roo Code agent is stored in the .roo/mcp.json
file within the VS Code workspace. Because the MCP configuration format allows for execution of arbitrary commands, prior to version 3.20.3, it would have been possible for an attacker with access to craft a prompt to ask the agent to write a malicious command to the MCP configuration file. If the user had opted-in to auto-approving file writes within the project, this would have led to arbitrary command execution. This issue is of moderate severity, since it requires the attacker to already be able to submit prompts to the agent (for instance through a prompt injection attack), for the user to have MCP enabled (on by default), and for the user to have enabled auto-approved file writes (off by default). Version 3.20.3 fixes the issue by adding an additional layer of opt-in configuration for auto-approving writing to Roos configuration files, including all files within the .roo/
folder.
The product constructs all or part of a command using externally-influenced input from an upstream component, but it does not neutralize or incorrectly neutralizes special elements that could modify the intended command when it is sent to a downstream component.
Command injection vulnerabilities typically occur when:
Many protocols and products have their own custom command language. While OS or shell command strings are frequently discovered and targeted, developers may not realize that these other command languages might also be vulnerable to attacks. Command injection is a common problem with wrapper programs.