Detection rules › Elastic

Potential cPanel WHM CRLF Authentication Bypass (CVE-2026-41940)

Status
production
Severity
high
Time window
9m
Author
Elastic
Source
github.com/elastic/detection-rules

Identifies the network signature of CVE-2026-41940, a pre-auth root-level authentication bypass in cPanel and WebHost Manager (WHM) caused by a CRLF injection in the session writer. The exploit-inherent shape on the wire is a GET / request to a cPanel/WHM admin port (typically TCP/2087, 2086, 2083, 2082, 2095, 2096) carrying an Authorization: Basic header whose base64-decoded value contains CRLF-injected session fields, which causes cpsrvd to respond with a 3xx redirect whose Location header leaks a /cpsessNNNNNNNNNN token granting the attacker a privileged session. This is the network-layer equivalent of the cPanel access_log artifact identified by Unfold and watchTowr as the first bulletproof detection for this CVE: a GET / recorded with auth_method=b (HTTP Basic). Legitimate access to GET / on a WHM admin port returns 200 with the login screen and never includes HTTP Basic credentials, so this combination is not produced by normal use.

MITRE ATT&CK coverage

TacticTechniques
Initial AccessT1190 Exploit Public-Facing Application

Rule body elastic

[metadata]
creation_date = "2026/05/07"
integration = ["network_traffic", "zeek"]
maturity = "production"
updated_date = "2026/05/28"

[rule]
author = ["Elastic"]
description = """
Identifies the network signature of CVE-2026-41940, a pre-auth root-level authentication bypass in cPanel and WebHost
Manager (WHM) caused by a CRLF injection in the session writer. The exploit-inherent shape on the wire is a `GET /`
request to a cPanel/WHM admin port (typically TCP/2087, 2086, 2083, 2082, 2095, 2096) carrying an
`Authorization: Basic` header whose base64-decoded value contains CRLF-injected session fields, which causes cpsrvd
to respond with a 3xx redirect whose `Location` header leaks a `/cpsessNNNNNNNNNN` token granting the attacker a
privileged session. This is the network-layer equivalent of the cPanel `access_log` artifact identified by Unfold and
watchTowr as the first bulletproof detection for this CVE: a `GET /` recorded with `auth_method=b` (HTTP Basic).
Legitimate access to `GET /` on a WHM admin port returns 200 with the login screen and never includes HTTP Basic
credentials, so this combination is not produced by normal use.
"""
false_positives = [
    """
    Authorized vulnerability scanners (Nessus, Tenable, Qualys, etc.) running CVE-2026-41940 plugins will reproduce the
    exploit shape. Validate against scan windows and source IPs of approved scanners before escalating.
    """,
]
from = "now-9m"
index = ["packetbeat-*", "logs-network_traffic.http*", "logs-zeek.http*"]
language = "lucene"
license = "Elastic License v2"
name = "Potential cPanel WHM CRLF Authentication Bypass (CVE-2026-41940)"
note = """## Triage and analysis

> **Disclaimer**:
> This investigation guide was created using generative AI technology and has been reviewed to improve its accuracy and relevance. While every effort has been made to ensure its quality, we recommend validating the content and adapting it to suit your specific environment and operational needs.

### Investigating Potential cPanel WHM CRLF Authentication Bypass (CVE-2026-41940)

CVE-2026-41940 is a critical (CVSS 9.8) authentication bypass in cPanel & WHM that gives an unauthenticated attacker
a root-privileged session on the host. The exploit chains a CRLF injection in the session writer with an
encryption-skip triggered by a malformed cookie, then uses how cPanel caches sessions to promote the injected data
into a privileged login. Around 1.5M cPanel instances were exposed at disclosure and exploitation has been observed in
the wild since 2026-02-23, two months before the patch.

This rule fires on the Stage 2 request/response shape: a `GET /` to a cPanel admin port that carries an
`Authorization: Basic` header and receives a 3xx redirect whose `Location` points at a freshly minted
`/cpsessNNNNNNNNNN` path. Per the watchTowr and Unfold writeups, this is the only request shape that lets the
exploit retrieve the security token needed for Stage 4 (privileged use of the session).

### Detection logic

The rule requires all of the following on a single decoded HTTP transaction matched from
`data_stream.dataset:network_traffic.http` (or `event.category:network_traffic` with `network.protocol:http`):

- `http.request.method:GET` and `url.path:"/"` — request targets the root path exactly. The CRLF vulnerability is only
  reachable on `GET /`; the same payload on other paths does not return the redirect that leaks the token, so the
  match is intentionally exact (a request like `GET /index.html` will not fire).
- `destination.port:(2087 OR 2086 OR 2083 OR 2082 OR 2095 OR 2096)` — cPanel/WHM admin and webmail ports. These are
  not in the default Network Packet Capture HTTP port list and must be added explicitly (see Setup).
- `http.response.status_code:[300 TO 399]` — a redirect response. Normal `GET /` to WHM returns 200 with the login
  screen; only the exploit produces a 3xx here.
- `http.request.headers.authorization:Basic*` — HTTP Basic credentials sent on `GET /`. This is the network-layer
  equivalent of the `auth_method=b` flag the Unfold and watchTowr writeups identify as the first bulletproof artifact
  in cPanel's `access_log`. `GET /` is an unauthenticated endpoint in normal cPanel operation and never legitimately
  carries Basic auth.
- `http.response.headers.location:/cpsess*` — the response redirects to a `/cpsess`-prefixed path, leaking the CSRF
  token the attacker needs for Stage 4. This is what makes the exploit succeed and is not produced by any benign flow.

### Possible investigation steps

- Capture the alert evidence. Record `source.ip` (attacker), `destination.ip` (cPanel host), `destination.port`,
  `user_agent.original`, `http.response.status_code`, the exact `http.response.headers.location` value (which contains
  the leaked `cpsess` token), and the captured `http.request.headers.authorization` value.
- Decode the Authorization header to confirm the CRLF payload. Strip the leading `Basic ` from
  `http.request.headers.authorization` and base64-decode the remainder. A legitimate Basic credential decodes to
  `username:password`; the exploit's payload decodes to a multi-line block delimited by `\\r\\n` containing fields like
  `successful_internal_auth_with_timestamp=`, `tfa_verified=1`, and `hasroot=1`. CRLF bytes in the decoded value
  distinguish exploitation from a misconfigured Basic-auth client.
- Confirm the destination host runs cPanel/WHM. Identify the installed version and whether the 2026-04-28 emergency
  patch is applied.
- Pivot on the source IP across the host's `/usr/local/cpanel/logs/access_log`. The exploit-inherent log artifact is a
  request line of the form `"GET / HTTP/1.1" 3xx 0 "-" "<UA>" "b" "-" <port>` — `auth_method=b` on `GET /` should never
  occur in normal operation and corresponds 1:1 to the `http.request.headers.authorization:Basic*` clause in this rule.
- Look for the Stage 4 follow-on from the same source IP: a request to the leaked `cpsess` path
  (`/cpsessNNNNNNNNNN/...`) with `auth_method=s` (session) and HTTP 200, without a preceding successful login
  (form POST `/login`, `/openid_connect/`, or reseller `?session=`). This is the post-exploitation artifact.
- Identify whether privileged WHM API actions were invoked under the leaked `cpsess` token (account creation, package
  install, file manager writes, terminal API).
- Review egress from the host for outbound connections initiated after the alert that could indicate web shell or
  implant install.

### False positive analysis

- Legitimate WHM administration never produces `GET /` with HTTP Basic authentication and a 3xx redirect leaking a
  fresh `cpsess` token. This combination is exploit-inherent.
- Authorized vulnerability scans running CVE-2026-41940 plugins will reproduce the request shape.

### Response and remediation

- Apply the cPanel emergency patch released 2026-04-28 (or the WP Squared equivalent). Verify by checking the
  installed cPanel version against the advisory.
- If the alert is paired with an `auth_method=s` `cpsess` request (post-exploitation), assume host compromise:
  rotate root credentials, audit `/var/cpanel/sessions/`, look for newly created accounts, scheduled tasks, SSH keys,
  and `authorized_keys` modifications.
- Restrict access to cPanel admin ports (2087/2086/2083/2082/2095/2096) to known administrator source IPs at the
  perimeter or via host firewall.
- Block the source IP at the WAF or perimeter if exploitation is confirmed.
"""
references = [
    "https://www.unfold.ai/blog/cpanel-exploit-cve-2026-41940",
    "https://labs.watchtowr.com/the-internet-is-falling-down-falling-down-falling-down-cpanel-whm-authentication-bypass-cve-2026-41940/",
    "https://www.picussecurity.com/resource/blog/cve-2026-41940-explained-cpanel-whm-authentication-bypass-hit-1-5m-servers",
    "https://support.cpanel.net/hc/en-us/articles/40073787579671-Security-CVE-2026-41940-cPanel-WHM-WP2-Security-Update-04-28-2026",
    "https://nvd.nist.gov/vuln/detail/CVE-2026-41940",
    "https://docs.cpanel.net/knowledge-base/cpanel-product/the-cpanel-log-files",
    "https://www.elastic.co/guide/en/beats/packetbeat/current/packetbeat-http-options.html#_send_all_headers",
]
risk_score = 73
rule_id = "2449be9d-2fdf-4126-a85b-f05e4058df9f"
setup = """## Setup

This rule supports two data sources:

1. **Network Packet Capture / Packetbeat (preferred):** Requires the Network Packet Capture integration (or legacy
   Packetbeat) with cPanel admin ports added to the HTTP protocol configuration and `send_all_headers` enabled, so that
   `http.request.headers.authorization` and `http.response.headers.location` are populated. cPanel admin ports
   (2087/2086/2083/2082/2095/2096) are not in the default HTTP port list and must be added explicitly.
2. **Zeek HTTP logs (`zeek.http`):** Zeek records only the names of HTTP request and response headers in
   `zeek.http.client_header_names` and `zeek.http.server_header_names`, not their values. The Zeek branch of this rule
   therefore matches on the presence of `AUTHORIZATION` and `LOCATION` headers on a `GET /` to a cPanel admin port,
   which is a less precise but still exploit-shaped signal. Confirm exploitation by retrieving the raw header values
   from the upstream Zeek sensor or correlated packet capture.

   The Zeek branch has non-default sensor prerequisites that must be met or it will never match:
   - **Header-name logging is not enabled by default.** `client_header_names` and `server_header_names` are produced by
     the policy script `policy/protocols/http/header-names.zeek`, which is not loaded by `base/protocols/http` or the
     default `site/local.zeek`. Add `@load policy/protocols/http/header-names.zeek` to the Zeek configuration. Without
     it, neither field is emitted and the Zeek branch cannot fire.
   - **Server header-name logging defaults to off.** Even with the script loaded, `HTTP::log_server_header_names`
     defaults to `F`, so the `server_header_names:LOCATION` condition is unsatisfiable. Set
     `redef HTTP::log_server_header_names = T;` (client header-name logging is on by default).
   - **Non-standard ports are handled by DPD, not a port list.** Zeek ships port-independent HTTP DPD signatures
     (`base/protocols/http/dpd.sig`) that detect HTTP by payload regardless of port, so plaintext `GET /` on cPanel
     admin ports (2086/2082/2095) is parsed into `zeek.http` without registering those ports. The TLS ports
     (2087/2083/2096) are encrypted and yield `zeek.ssl`, not `zeek.http`, unless the sensor sits upstream of TLS
     termination (see decryption note below).

cPanel/WHM exposes paired TLS and plaintext ports: 2087 (WHM HTTPS), 2083 (cPanel HTTPS), and 2096 (Webmail HTTPS)
require decryption visibility (TLS interception, sidecar on the host, or a sensor upstream of TLS termination) for
either data source to observe HTTP headers. The plaintext counterparts — 2086 (WHM HTTP), 2082 (cPanel HTTP), and 2095
(Webmail HTTP) — carry headers in the clear and are observable without decryption. An attacker can exploit the
vulnerability over either variant, so both sets of ports are included in this rule.
"""
severity = "high"
tags = [
    "Domain: Network",
    "Domain: Application",
    "Domain: Web",
    "Use Case: Threat Detection",
    "Use Case: Vulnerability",
    "Tactic: Initial Access",
    "Data Source: Network Packet Capture",
    "Data Source: Network Traffic",
    "Data Source: Zeek",
    "Resources: Investigation Guide",
]
timestamp_override = "event.ingested"
type = "query"

query = '''
(
  (
    (data_stream.dataset:network_traffic.http OR (event.category:network_traffic AND network.protocol:http)) AND
    http.response.status_code:[300 TO 399] AND
    http.request.headers.authorization:Basic* AND
    http.response.headers.location:\/cpsess*
  )
  OR
  (
    data_stream.dataset:zeek.http AND
    zeek.http.client_header_names:AUTHORIZATION AND
    zeek.http.server_header_names:LOCATION
  )
) AND
http.request.method:GET AND
url.path:"/" AND
destination.port:(2087 OR 2086 OR 2083 OR 2082 OR 2095 OR 2096)
'''


[[rule.threat]]
framework = "MITRE ATT&CK"

[[rule.threat.technique]]
id = "T1190"
name = "Exploit Public-Facing Application"
reference = "https://attack.mitre.org/techniques/T1190/"

[rule.threat.tactic]
id = "TA0001"
name = "Initial Access"
reference = "https://attack.mitre.org/tactics/TA0001/"

Stages and Predicates

Stage 1: query

(
  (
    (data_stream.dataset:network_traffic.http OR (event.category:network_traffic AND network.protocol:http)) AND
    http.response.status_code:[300 TO 399] AND
    http.request.headers.authorization:Basic* AND
    http.response.headers.location:\/cpsess*
  )
  OR
  (
    data_stream.dataset:zeek.http AND
    zeek.http.client_header_names:AUTHORIZATION AND
    zeek.http.server_header_names:LOCATION
  )
) AND
http.request.method:GET AND
url.path:"/" AND
destination.port:(2087 OR 2086 OR 2083 OR 2082 OR 2095 OR 2096)

Indicators

Each row is a field, operator, and value that the rule matches. The corpus column counts how many other rules in the catalog look for the same combination: high numbers point to widely-used, community-vetted indicators. Blank or 1 shows that the indicator is specific to this rule.