CVE WatchJul 15, 2026

CVE-2026-42271

BerriAI LiteLLM Command Injection Vulnerability

CVSS

8.8

High

EPSS

92.6%

p100

KEV

YES

Jun 8, 2026

Exploit Today

0-100

Published: May 8, 2026 · Last modified: Jul 15, 2026 · CWE-77 · CWE-78

EPSS · 30d

12.8%EPSS · 30 days92.6%

2026-09-162026-10-14

KEV catalog record

Product

BerriAI / LiteLLM

Vulnerability

BerriAI LiteLLM Command Injection Vulnerability

Added to KEV

Jun 8, 2026

Remediate by

Jun 22, 2026

Known ransomware use

Summary description

BerriAI LiteLLM contains a command injection vulnerability that could allow any authenticated user, including holders of low-privilege internal-user keys, to run arbitrary commands on the host.

Required action

Apply mitigations per vendor instructions, follow applicable BOD 22-01 guidance for cloud services, or discontinue use of the product if mitigations are unavailable.

Notes

This vulnerability affects a common open-source component, third-party library, or a protocol used by different products. Please check with specific vendors for information on patching status. For more information, please see: https://github.com/BerriAI/litellm/security/advisories/GHSA-v4p8-mg3p-g94g ; https://github.com/BerriAI/litellm/releases/tag/v1.83.7-stable ; https://nvd.nist.gov/vuln/detail/CVE-2026-42271

Technical description

LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format. From version 1.74.2 to before version 1.83.7, two endpoints used to preview an MCP server before saving it — POST /mcp-rest/test/connection and POST /mcp-rest/test/tools/list — accepted a full server configuration in the request body, including the command, args, and env fields used by the stdio transport. When called with a stdio configuration, the endpoints attempted to connect, which spawned the supplied command as a subprocess on the proxy host with the privileges of the proxy process. The endpoints were gated only by a valid proxy API key, with no role check. Any authenticated user — including holders of low-privilege internal-user keys — could therefore run arbitrary commands on the host. This issue has been patched in version 1.83.7.

Official references

Related CVEs

CVECVSSEPSSKEVRExploitTitleMod.

CVE-2026-1112297.2 HIG

—

———pfSense CE 2.7.0 through 2.9.0 contains a command injection vulnerability in services_captiveportal_hasync.php that allows users with only the Captive Portal HA privilege to execute commands as root. Attackers can set backwardsyncip to a malicious XMLRPC server returning connected_users with shell metacharacters in ip or mac fields, which captiveportal_load_pfctl passes unescaped to a pfctl shell command.1h CVE-2026-1112267.2 HIG

—

———pfSense CE 2.4.5 through 2.9.0 contains an OS command injection vulnerability that allows authenticated WebGUI users with Advanced Admin Access privilege to execute commands as root via Login Protection settings. Attackers can submit sshguard threshold, block time, or detection time values containing newlines or $(...) that are written unvalidated to sshguard.conf and sourced as root on restart.1h CVE-2026-1112255.0 MED

—

———Netgate pfSense CE 2.5.0 through 2.9.0 contains an OS command injection vulnerability in openvpn.learn-address.sh that allows authenticated VPN clients to execute commands as the unbound user. Attackers can embed shell metacharacters in the certificate common name or username, which is interpolated unquoted into su -c strings, when DNS Resolver client registration is enabled.1h CVE-2026-1112247.5 HIG

—

———pfSense CE through 2.9.0 contains an OS command injection vulnerability in the OpenVPN openvpn.connect_async.sh script that allows authenticated VPN users to execute commands as root via crafted usernames. Attackers authenticated through RADIUS or LDAP can use usernames containing single quotes and shell metacharacters to break out of the eval'd pfctl anchor name on connect or disconnect.1h CVE-2026-1112187.2 HIG

—

———Netgate pfSense CE 2.7.0 through 2.9.0 contains an OS command injection vulnerability that allows authenticated users with Traffic Shaper Limiters privileges to execute root commands by tampering with the limiter number field. Attackers can submit a malicious hidden 'number' value when editing a limiter, then delete it so delete_queue() passes the unquoted value to dnctl via /bin/sh as root.1h CVE-2026-1112177.2 HIG

—

———Netgate pfSense CE through 2.9.0 contains an OS command injection vulnerability that allows authenticated users with Virtual IP edit privileges to execute commands as root by storing an unvalidated CARP VHID. Attackers can supply a VHID containing shell metacharacters that executes via interface_ipalias_configure() ifconfig calls or get_carp_interface_status() when CARP status is rendered.1h