The MCP Inspector proxy trusted a redirect and inherited command execution
The MCP Inspector proxy trusted a redirect and inherited command execution Local developer tools sit in a useful but dangerous position. They run on a developer workstation, can often launch processes, and connect to servers supplied by the same user. MCP Inspector was built to test those servers, so a hostile server was always part of its threat model. The advisory GHSA-g9hg-qhmf-q45m describes…
The MCP Inspector proxy was found to trust a redirect and inherit command execution in a vulnerability described by CVE-2025-58444. This issue, classified as high severity under CVSS v4, allows a malicious redirect URI from an untrusted remote MCP server to inject script into the Inspector interface. The proxy then provides a pathway for this script execution to execute arbitrary commands on the developer's machine.
This is due to the Inspector rendering the redirect URI in a context where it could become executable script, a classification that falls under CWE-84, improper neutralization of encoded URI schemes in a web page. The impact is high, as the resulting process could potentially gain access to sensitive developer resources such as repository access, cloud tokens, SSH keys, package registry credentials, and shell history.
The vulnerability is addressed in version 0.16.6 of @modelcontextprotocol/inspector. To mitigate the risk, developers are advised to upgrade to this version and review how the Inspector is launched, ensuring it is configured to listen only on loopback interfaces and requiring authentication for remote boundaries.
Brief written by urgent.news; may contain errors — check the original.
Source: dev.to