Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB
CVE-2026-114725 Overview
CVE-2026-114725 is a mail header injection vulnerability affecting Roundcube Webmail versions before 1.6.20 and 1.7.x before 1.7.5. The rcmail_sendmail function stripped newline characters only from the Subject and Organization headers. Other header values taken from request parameters remained unsanitized. A remote authenticated attacker can inject arbitrary headers into outgoing messages through the References, In-Reply-To, Message-ID, and From parameters. The flaw is classified under [CWE-99] (Improper Control of Resource Identifiers).
Critical Impact
Authenticated users can manipulate outgoing SMTP headers, enabling message spoofing, routing manipulation, and abuse of mail infrastructure for spam or phishing campaigns.
Affected Products
- Roundcube Webmail versions prior to 1.6.20
- Roundcube Webmail 1.7.x versions prior to 1.7.5
- Downstream distributions bundling vulnerable Roundcube builds (see Debian Bug Report #1151279)
Discovery Timeline
- 2026-10-19 - Roundcube publishes Security Updates Overview for 1.6.20 and 1.7.5
- 2026-10-22 - CVE-2026-114725 published to NVD
- 2026-10-22 - Last updated in NVD database
Technical Details for CVE-2026-114725
Vulnerability Analysis
The vulnerability resides in program/include/rcmail_sendmail.php, which handles outgoing message composition in Roundcube Webmail. The sanitization logic enforced a CRLF-stripping routine solely against the Subject header value and, in related paths, the Organization header. Header parameters sourced from user input, including References, In-Reply-To, Message-ID, and From, were passed directly to the mail backend without whitespace normalization. An authenticated attacker can embed CR or LF sequences in these parameters to append additional headers or inject new message content into the outgoing SMTP transaction.
Root Cause
The root cause is incomplete input neutralization. The original sanitizer treated newline removal as a header-specific concern rather than a transport-level invariant. The upstream fix, titled "Fix mail header injection via References/In-Reply-To/Message-ID/From parameters," normalizes whitespace in every string header value instead of enumerating specific headers for sanitization.
Attack Vector
Exploitation requires an authenticated session on the Roundcube webmail interface. The attacker submits a message compose request that contains CRLF byte sequences inside one of the vulnerable parameters. The injected content is accepted by rcmail_sendmail, which forwards it to the SMTP layer. Common abuse scenarios include forging Bcc recipients, overriding Return-Path, inserting routing headers to bypass filters, or attaching crafted MIME bodies for phishing payload delivery.
$message_id = $this->rcmail->gen_message_id($from);
}
- // Don't allow CRLF in subject (#8404)
- $subject = trim(preg_replace('/[\r\n]+/', ' ', $subject));
-
$this->options['dsn_enabled'] = $dsn_enabled;
$this->options['from'] = $from;
$this->options['mailto'] = $mailto;
Source: Roundcube Commit #79adcfc0. The patch removes the narrow subject-only CRLF filter and relocates whitespace normalization to a universal header handling routine. A parallel change was applied in Roundcube Commit #acd32a86 for the 1.7.x branch.
Detection Methods for CVE-2026-114725
Indicators of Compromise
- Outgoing messages originating from Roundcube with duplicated or anomalous Message-ID, References, or In-Reply-To headers.
- SMTP submission events containing embedded CR (%0D) or LF (%0A) byte sequences in POST parameters to the compose endpoint.
- Unexpected Bcc, Cc, or Return-Path headers appearing in mail logs for messages sent by low-privilege Roundcube accounts.
Detection Strategies
- Inspect web server access logs for POST requests to ?_task=mail&_action=send that include URL-encoded CRLF sequences in body parameters.
- Correlate authenticated Roundcube session identifiers with mail transfer agent logs to identify users emitting messages with malformed header blocks.
- Deploy web application firewall rules that reject requests containing \r\n characters within Roundcube compose parameters.
Monitoring Recommendations
- Enable verbose SMTP submission logging on the mail transport agent (Postfix, Exim, or Sendmail) to retain full header sets for post-incident review.
- Alert on sudden volume increases in outbound mail per authenticated Roundcube user.
- Track reputation signals and bounce traffic to detect abuse campaigns leveraging compromised accounts.
How to Mitigate CVE-2026-114725
Immediate Actions Required
- Upgrade to Roundcube Release 1.6.20 or Roundcube Release 1.7.5 as applicable to your deployment branch.
- Audit mail transfer agent logs for header anomalies produced since the vulnerable Roundcube build was deployed.
- Rotate credentials for Roundcube accounts suspected of abuse and enforce multi-factor authentication where supported.
Patch Information
The fix is delivered through Roundcube Commit #79adcfc0 and Roundcube Commit #acd32a86. Both patches apply whitespace normalization to every string header value processed by rcmail_sendmail. Operators relying on Debian packages should track the status noted in Debian Bug Report #1151279.
Workarounds
- Restrict Roundcube submission to trusted networks using reverse proxy access control until the patched release can be deployed.
- Configure the upstream SMTP relay to reject messages containing bare CR or LF characters inside header fields.
- Apply rate limiting on the compose endpoint to constrain abuse by compromised credentials.
# Example ModSecurity rule blocking CRLF injection in Roundcube compose parameters
SecRule REQUEST_URI "@contains _task=mail" \
"id:1142501,phase:2,deny,status:400,msg:'Possible Roundcube header injection (CVE-2026-114725)',\
chain"
SecRule ARGS:_references|ARGS:_in_reply_to|ARGS:_message_id|ARGS:_from "@rx (?:%0[aAdD]|\r|\n)"
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB

