This is a loaded nail gun left behind a locked maintenance door
CVE-2026-11413 is a stack-based buffer overflow in the set_macfilter function inside /sbin/jdcweb_rpc on the JingDong JD Cloud Box AX6600, disclosed on 2026-06-06. The record and public mirrors describe the affected target as firmware 4.5.3.r4546 and score it with PR:L, which means the published attack model assumes the attacker already has a valid low-privilege account on the device's management surface.
The vendor-style 8.8/HIGH score is technically defensible in a lab because an authenticated network attacker can plausibly drive memory corruption to full device compromise. In enterprise reality, that overstates urgency: this is a niche consumer/SMB router, typically LAN-side only, and the attack precondition is already a meaningful compromise stage. Public exploit code keeps it out of backlog-junk territory, but the attacker-position requirement and limited enterprise population pull it down to MEDIUM.
4 steps from start to impact.
Reach the AX6600 management plane
jdcweb_rpc. In most real deployments this means local LAN presence, VPN access, or prior foothold inside a branch/home-office segment rather than blind internet exposure.- JD Cloud Box AX6600 deployed
- Management interface reachable from the attacker's network position
- Target running firmware 4.5.3.r4546 or another build with the vulnerable
set_macfilterimplementation
- These devices are not common enterprise standard kit
- Router admin planes are often bound to RFC1918 space only
- NGFW/VPN segmentation frequently blocks direct management access from untrusted zones
Authenticate with low privileges
PR:L, so the exploit chain assumes valid credentials before hitting the vulnerable code path. That makes this a post-auth bug, not an unauthenticated perimeter-breaker.- Valid low-privilege account or equivalent authenticated session
- Authentication endpoint exposed and usable by the attacker
- Requires credential theft, password reuse, insider access, or a prior device compromise
- MFA is uncommon on this class of device, but admin credential hygiene still blocks many casual attacks
Trigger set_macfilter overflow with public exploit tooling
set_macfilter function in /sbin/jdcweb_rpc, causing a stack-based buffer overflow. NVD and Tenable both reference a public exploit archive, so weaponization effort is low once access and credentials exist.- Authenticated access to the vulnerable API path
- Payload length/control sufficient to corrupt stack state
- Reliable code execution from a crash-prone stack overflow on embedded MIPS/ARM firmware can still require target-specific tuning
- Some attempts may only crash the service or device, creating noise before yielding code execution
Take over the router and pivot
- Successful memory-corruption exploitation beyond simple crash
- Router privileges sufficient to change config and observe traffic
- Impact is bounded to organizations that actually use this hardware
- Compromise of one branch router is serious, but not automatically domain-wide
The supporting signals.
| In-the-wild status | No authoritative evidence of broad in-the-wild exploitation found in the sources reviewed. Not in CISA KEV as of the accessible catalog view, and no public campaign reporting was surfaced in GreyNoise/Censys-accessible search results. |
|---|---|
| Proof-of-concept availability | Yes. NVD and Tenable both reference a public exploit archive: JDcloud-AX6600_overflow.zip. That sharply lowers attacker effort once prerequisites are met. |
| EPSS | No EPSS value reliably observed yet in accessible sources. Given the disclosure date of 2026-06-06, this is likely too new for downstream mirrors to show stable scoring; that is an inference from source timing, not a FIRST-confirmed zero. |
| KEV status | Not KEV-listed. CISA's Known Exploited Vulnerabilities Catalog is the authoritative source and does not show this CVE in the reviewed material. |
| CVSS vector and what it really means | CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H = remote and low complexity, but authenticated. PR:L is the whole story: this is not a clean perimeter smash; it assumes the attacker already got onto the management plane with credentials. |
| Affected versions | Published records name JingDong JD Cloud Box AX6600 firmware 4.5.3.r4546 specifically. There is no authoritative broad version range in the reviewed sources beyond that cited build. |
| Fixed version | No vendor-fixed firmware version located in the reviewed sources. Unlike some earlier AX6600 issues, this record does not currently point to a patch advisory or confirmed remediated build. |
| Scanning and exposure data | I did not find reliable Shodan/Censys/FOFA counts tied to this exact CVE or build in accessible primary sources. That usually means defenders should assume unknown but probably limited enterprise prevalence, not safety. |
| Disclosure timeline | NVD shows Published: 2026-06-06 and notes the record was received from VulDB on 2026-06-06. |
| Reporter / source of record | The CNA/source shown by NVD is VulDB. Public references include VulDB's entry and submission page for the flaw, with the issue framed as a stack overflow in set_macfilter. |
noisgate verdict.
The decisive factor is PR:L on a router management interface: this bug pays off only after the attacker already has authenticated access to a niche appliance. Public exploit code keeps the risk real, but the reachable population and attacker-position requirement are both materially narrower than the vendor-style 8.8 implies.
Why this verdict
- Down from 8.8 because
PR:Lmeans post-auth: the attacker needs valid credentials or an authenticated session before touching the vulnerable path. - Down again because exposure is narrow: this is a JD Cloud consumer/SMB router, not a mainstream enterprise platform deployed across thousands of corporate nodes.
- Down again because management-plane reachability is usually limited: in modern deployments the admin UI tends to sit on LAN/VPN space, not broad internet exposure.
- Back up slightly because a public exploit exists: once an attacker has access, the barrier to testing and weaponization is lower than a source-only or theoretical memory-corruption bug.
- Back up slightly because router compromise is strategically useful: a successful hit can alter DNS, firewalling, and traffic visibility at the edge.
Why not higher?
This does not deserve HIGH or CRITICAL unless you know these routers are broadly deployed and their management planes are reachable by attackers at scale. The published vector's PR:L requirement is a hard reality check: authenticated remote on an edge appliance is materially less urgent than unauthenticated internet RCE.
Why not lower?
This is not backlog fluff because the flaw is memory corruption on a router control binary with public exploit material already referenced by NVD. If you do run these devices, a compromised low-privilege account can plausibly become full device takeover, which is too dangerous for a LOW/IGNORE bucket.
What to do — in priority order.
- Restrict management-plane reachability — Move the AX6600 admin surface behind VPN, dedicated management VLANs, or host-based ACLs so only trusted admin stations can reach it. For a MEDIUM verdict there is no noisgate mitigation SLA, so do this at the next practical change window while you track firmware remediation.
- Rotate and minimize router accounts — Because the exploit path assumes
PR:L, kill shared credentials, remove stale users, and rotate admin and support passwords now. There is no noisgate mitigation SLA for MEDIUM, but this directly attacks the most important prerequisite and should not wait for firmware news. - Watch for router config drift — Alert on DNS, firewall, WAN-management, and account changes on these devices, plus unexpected reboots of
jdcweb_rpc-adjacent services. Use this as a detection backstop until you can verify patched firmware and complete remediation within the remediation window. - Segment branch and home-office edge devices — Treat these routers as semi-trusted appliances and prevent them from reaching core management networks except where explicitly needed. Again, no noisgate mitigation SLA applies at MEDIUM, but segmentation cuts blast radius immediately.
- A perimeter WAF does not materially help if the vulnerable endpoint is on a local router management service rather than a web app behind your reverse proxy.
- Standard endpoint EDR on user laptops does not protect the router itself; the vulnerable component is the appliance binary
/sbin/jdcweb_rpc. - Relying on 'not internet-facing' alone is weak control because the bug is post-auth; any insider, VPN user, or attacker with prior foothold can still reach it.
Crowdsourced verification payload.
Run this on the target AX6600 over SSH or local shell, ideally as root or an admin-capable shell user with read access to /sbin/jdcweb_rpc and firmware metadata. Example: sh check_cve_2026_11413.sh. It checks whether the device appears to be an AX6600 on firmware 4.5.3.r4546 and whether the vulnerable binary exposes set_macfilter; because no authoritative fixed version was found, anything else is reported as PATCHED only relative to the published affected build, not as a vendor-confirmed safe version.
#!/bin/sh
# check_cve_2026_11413.sh
# Detect likely exposure to CVE-2026-11413 on JingDong JD Cloud Box AX6600
# Exit codes: 0=PATCHED, 1=VULNERABLE, 2=UNKNOWN
set -u
get_version() {
for f in /etc/openwrt_release /etc/os-release /etc/version /proc/version; do
[ -r "$f" ] || continue
v=$(grep -Eo '4\.5\.3\.r4546|4\.5\.[0-9]+\.r[0-9]+' "$f" 2>/dev/null | head -n 1)
if [ -n "$v" ]; then
echo "$v"
return 0
fi
done
return 1
}
get_model() {
for f in /tmp/sysinfo/model /proc/device-tree/model /etc/board.json /etc/openwrt_release; do
[ -r "$f" ] || continue
m=$(strings "$f" 2>/dev/null | grep -Ei 'AX6600|JD Cloud Box|JingDong|JDCloud' | head -n 1)
if [ -n "$m" ]; then
echo "$m"
return 0
fi
done
return 1
}
BIN="/sbin/jdcweb_rpc"
MODEL="$(get_model || true)"
VERSION="$(get_version || true)"
HAS_BIN=0
HAS_FUNC=0
if [ -x "$BIN" ] || [ -r "$BIN" ]; then
HAS_BIN=1
if strings "$BIN" 2>/dev/null | grep -q 'set_macfilter'; then
HAS_FUNC=1
fi
fi
if [ -z "$MODEL" ] && [ "$HAS_BIN" -eq 0 ]; then
echo "UNKNOWN - unable to confirm AX6600 model or locate $BIN"
exit 2
fi
if [ "$HAS_BIN" -eq 1 ] && [ "$HAS_FUNC" -eq 1 ] && [ "$VERSION" = "4.5.3.r4546" ]; then
echo "VULNERABLE - AX6600 indicators present, $BIN exports set_macfilter, firmware appears to be 4.5.3.r4546"
exit 1
fi
if [ -n "$VERSION" ] && [ "$VERSION" != "4.5.3.r4546" ]; then
echo "PATCHED - firmware is not the specifically published affected build (found: $VERSION). Note: no vendor-confirmed fixed version was available in reviewed sources."
exit 0
fi
if [ "$HAS_BIN" -eq 1 ] && [ "$HAS_FUNC" -eq 0 ]; then
echo "PATCHED - $BIN present but set_macfilter symbol/string not observed"
exit 0
fi
echo "UNKNOWN - insufficient evidence to map this device to the published vulnerable build"
exit 2
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.