Skip to main content
Mallory
HighExploited in the wildCISA KEVPublic exploitVulnerability

CVE-2026-114713: Server-to-server command injection in UnrealIRCd labeled-response

Published Nov 8, 2026 · Updated today
Trending this week#5of 427,535 vulnerabilities
Mention activity · 30 days139 mentions · peak 51 in a dayToday
CVSS7.7High · v3.1
EPSS0.51%44th percentile
CISA KEVYesAdded Nov 13, 2026
In the wildActivePer CISA KEV
IdentifiersCVE-2026-114713CWE-140 · Improper Neutralization of…

CVE-2026-114713 is an improper delimiter neutralization vulnerability in the default-enabled labeled-response module of UnrealIRCd 6.0.0 through versions before 6.2.7.1. The module fails to escape the IRCv3 label message tag when echoing client-supplied values into replies. Crafted labels can introduce newlines and additional IRC protocol commands into server-to-server traffic, causing linked servers to process attacker-controlled commands with server authority. Exploitation requires a network containing at least two linked servers and is possible for an ordinary connected user without IRC operator privileges. Unsupported UnrealIRCd 5 installations are also affected.

Do you run this?See affected versions in your environment and whether adversaries are exploiting this vulnerability against assets like yours.
Check your exposure
Analyst brief

Impact, mitigation & remediation

What it means and what to do now: patch path, interim mitigations, and the assume-compromise checklist.

Impact

What an attacker gets, and what they’ve been doing with it.

An ordinary IRC user can cross the client-to-server trust boundary and perform privileged IRC operations with server or IRC operator authority. Consequences include disconnecting users through KILL commands, imposing network-wide GLINE bans, manipulating users and channels, and disrupting IRC network operations. The vulnerability enables IRC protocol command execution; it does not establish arbitrary operating-system command execution. Active exploitation has been reported.

Mitigation

If you can’t patch tonight, do this now.

If immediate patching is not possible and every linked server runs UnrealIRCd 6.0.4 or newer, deploy the vendor-provided SPAMFILTER workaround to block malformed label tags across the network. Alternatively, disable the labeled-response module in the configuration on every server and restart each server; a configuration rehash is insufficient. These measures do not replace patching. Review linked-server logs for unexpected KILL, GLINE, or other unauthorized administrative operations.

Remediation

Patch, then assume compromise.

Upgrade every affected hub and leaf server to UnrealIRCd 6.2.7.1 or a later supported release and restart the upgraded services. Alternatively, apply the vendor's bundle-20261107 hot-patch separately to each compatible Unix-like installation running UnrealIRCd 6.0.0 through 6.2.7. Verify that the labeled-response module reports the patched module version. Windows installations and systems where hot-patching fails must upgrade and restart. Migrate unsupported UnrealIRCd 5 installations to a fixed, supported release rather than relying on temporary workarounds.

Public exploits

Exploits

1 exploit validated by Mallory's analysis. Hidden: 2 rejected by Mallory's analysis as not a working exploit.

View more in app →
CVE-2026-114713-POCThe supplied repository contains a standalone Python exploit and a five-instance reproduction lab, not an exploit-framework module. Its 18 files comprise poc.py, two Bash setup scripts, a Dockerfile, Docker Compose orchestration, a sourced build-settings template, five IRC server configurations, five evidence transcripts, a README, and an MIT license. The four code files counted are the Python client, two Bash scripts, and Dockerfile; the settings template is treated as configuration. poc.py uses only Python standard-library modules. It registers three ordinary IRC clients, requests labeled-response and batch for the attacker, sends a direct KILL as a privilege-control attempt, and then submits a labeled two-parameter WHOIS. The hardcoded label contains escaped newline and space characters. According to the supplied explanation, vulnerable labeled-response formatting emits the decoded LF without re-escaping it, allowing a response crossing a trusted server link to become an additional IRC command. The implemented payload kills the test client 'vic'; there is no shell, persistence, exfiltration, or operating-system RCE payload. Broader server-command abuse, including bans, is discussed but not implemented. The documented decoded-label budget is 64 bytes. The lab compares linked UnrealIRCd 6.2.7 servers with linked 6.2.7.1 controls and an unlinked 6.2.7 control. All configured listeners bind loopback; Docker uses Linux host networking. Build scripts fetch upstream release archives and generate TLS material. Configuration files contain static lab operator credentials and a shared link password, but the exploit does not use them. Normal package-cache and downloaded-archive cleanup commands do not indicate a fake exploit. Supplied transcripts show direct KILL rejection, a server-attributed victim disconnect on the vulnerable linked topology, and surviving victims on both controls. These are repository-provided records, not independently executed verification. The verdict logic is weaker than its wording: it checks for KILL-related text and EOF but does not validate the claimed server's identity, and the direct privilege pre-check is printed rather than enforced. Failure to observe a kill is not conclusive proof of patching. Detection should correlate unexpected KILL events with the distinctive injected reason and appended BATCH protocol text rather than treat every server-originated KILL as malicious. The repository identifies CVE-2026-114713, claims affected versions 6.0.0–6.2.7 and 5.x, and documents 6.2.7.1 or hot-patch bundle-20261107 as remediation. Advisory, KEV, and exploitation-in-the-wild claims were not independently verified. No repository URL, git reference, or archive size was supplied; empty strings and zero represent unavailable repository metadata.fixedbyreboot · Disclosed Nov 14, 2026 · python · bash · networkOperational
FAQ

Frequently asked questions

Is CVE-2026-114713 exploited in the wild?
Yes. As of Nov 16, 2026, Mallory has reports of exploitation in the wild, and CISA added it to KEV on Nov 13, 2026.
Is there a public exploit for CVE-2026-114713?
Yes. Mallory tracks 1 public exploit or proof of concept for CVE-2026-114713, the earliest disclosed Nov 14, 2026.
Is there a patch for CVE-2026-114713?
Upgrade every affected hub and leaf server to UnrealIRCd 6.2.7.1 or a later supported release and restart the upgraded services. Alternatively, apply the vendor's bundle-20261107 hot-patch separately to each compatible Unix-like installation running UnrealIRCd 6.0.0 through 6.2.7.
How severe is CVE-2026-114713?
CVE-2026-114713 has a CVSS v3.1 score of 7.7 (High) and an EPSS score of 0.51% (44th percentile) as of Nov 16, 2026.
Activity feed

Recent activity

139 sources tracked across advisories, community write-ups, and news. New activity surfaces here as Mallory finds it.

View more in app →
cyberveilleNews · Nov 16, 2026🪲 Semaine 46 - CVE les plus discutées | CyberVeilleUnrealIRCd's labeled-response module fails to escape IRCv3 label tags, allowing users on networks with at least two linked servers to inject server-to-server commands executed with server authority. Versions from 6.0.0 before 6.2.7.1 are affected. The page reports CISA KEV inclusion, exploitation signals, and published proof-of-concept material.Read more ↗security weekNews · Nov 16, 2026Recent UnrealIRCd Vulnerability Exploited in Attacks - SecurityWeekA high-severity vulnerability in UnrealIRCd's labeled-response module, which fails to escape the IRCv3 label message tag in replies. On networks with two or more linked servers, an ordinary user can inject server-to-server IRC commands executed with server authority, including KILL and GLINE. Active probing and exploitation were reported on November 8, one day after disclosure. UnrealIRCd 6.2.7.1 and a hot patch address the issue; a spamfilter workaround is also available. UnrealIRCd 5 and earlier are end-of-life and will not receive fixes. CISA added the vulnerability to KEV on November 13.Read more ↗uganda certNews · Nov 16, 2026UnrealIRCd’s “Most Critical Issue in 15 Years” Exploited Within a Day of the Fix (CVE-2026-114713) | Uganda National Computer Emergency Response Team-National CertAn improper delimiter neutralisation vulnerability affecting UnrealIRCd versions before 6.2.7.1 allows attackers to inject and execute arbitrary commands with server or IRCOp-level privileges. The content assigns a CVSS score of 8.6 and reports exploitation attempts beginning November 8, 2026, one day after the patched release. Administrators are urged to update immediately; the stated CISA remediation deadline for federal agencies is November 16, 2026.Read more ↗risky biz rssNews · Nov 16, 2026Risky Bulletin: Hackers accessed Barracuda's cloudA command-injection vulnerability in UnrealIRCd that attackers are exploiting to take over IRC servers. Exploitation began one day after a patch was released.Read more ↗
+10 more in the timeline. Sign up to see them
Do you run this?See affected versions in your environment and whether adversaries are exploiting this vulnerability against assets like yours.
Check your exposure
Mallory platform

The version that knows your environment.

This page is what’s public. Mallory adds the parts that aren’t: which of your assets are affected, which adversaries are exploiting it right now, and which detections to deploy tonight.
Exposure mappingQuery your assets running an affected version, and investigate the blast radius.
Threat actor evidenceEvery report linking this CVE to a named adversary.
Associated malwareMalware families linked to this CVE, with evidence and IOCs.
Detection signatures1YARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Vendor-by-vendor mappingCross-references every affected SKU, including bundled OEM variants.
Social activity124Community discussion across Reddit, Mastodon, and other social sources.