BerriAI · LiteLLM · BerriAI LiteLLM Command Injection Vulnerability
Verdict
Patch soon · actively exploited, not reachable unauthenticated from the internet, and in the top 5% of CVEs by near-term exploitation probability
Added to CISA KEV 2026-06-08 · exploited in the wild for 111 days
CVSS 8.7 HIGH. The score is a summary; these three components are the part that decides whether something scanning the internet can use it.
| Attack vector | networkreachable over the network, with no local access needed |
|---|---|
| Privileges required | lowneeds some credentials, so a fully anonymous attacker cannot reach it directly |
| User interaction | nonecan be triggered with nobody interacting with the page or request |
| Base vector | CVSS:4.0/AV:N/AC:L/AT:P/PR:L/UI:N/VC:H/VI:H/VA:H/SC:H/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X |
| Weakness | CWE-77 |
This describes the vulnerability. It says nothing about whether your systems run it.
EPSS puts the probability of exploitation in the next 30 days at 0.1275, which places it in the top 5% of all tracked CVEs. The percentile is the more useful half: 0.18 in absolute terms sounds low until you know most of the internet-facing software in the world scores under that.
| EPSS probability | 0.1275 |
|---|---|
| Percentile | 0.961 — top 5% |
| Added to KEV | 2026-06-08 |
|---|---|
| Days exploited | 111 |
| Federal remediation due | not set |
| Ransomware use | no known ransomware campaign use |
Apply mitigations per vendor instructions, follow applicable BOD 22-01 guidance for cloud services, or discontinue use of the product if mitigations are unavailable.
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.