← Back to Feed CACHED · 2026-05-17 09:42:19 · cache_key CVE-2025-29912
CVE-2026-20180 · CWE-22 · Disclosed 2026-04-15

Cisco Identity Services Engine Multiple Remote Code Execution Vulnerability

ASSESSED — NOISGATE V0.5
Vendor
Reassessed
Verdict:
01 · The Real Story

This is not a stranger kicking in the front door, it is a badge holder finding a master key in the closet

CVE-2026-20180 is an authenticated remote code execution flaw in Cisco Identity Services Engine (ISE) caused by insufficient validation of user-supplied input in the web management interface. Cisco says an attacker with at least Read Only Admin credentials can send a crafted HTTP request, gain user-level access on the underlying OS, and then escalate to root. Cisco's fixed-version guidance says affected trains are earlier than 3.2, 3.2 before Patch 8, 3.3 before Patch 8, and 3.4 before Patch 4; 3.5 is not vulnerable. In single-node deployments, exploitation can also knock the node unavailable and block new endpoint authentications.

Cisco's 9.9 CRITICAL label is technically defensible on impact, but it overstates the *reachability* for enterprise prioritization. This is not unauthenticated internet RCE: the attacker needs management-plane reachability plus valid admin credentials, and Cisco specifically says Read Only Admin is enough. That makes this a dangerous post-compromise / credential-abuse amplifier against a high-value appliance, not a universal emergency on the same level as pre-auth edge-device RCE.

"Root on the NAC brain matters, but this starts with read-only admin on the ISE management plane, not the open internet."
02 · The Attack Path

4 steps from start to impact.

STEP 01

Reach the ISE management plane with curl or Burp Suite

The vulnerable surface is the web-based management interface on a Cisco ISE node. An attacker must be able to send HTTP requests to that admin plane, which usually means internal network access, VPN access, or a badly exposed management interface rather than raw internet reachability.
Conditions required:
  • Network path to the ISE web management interface
  • Target is running a vulnerable ISE release
Where this breaks in practice:
  • Many enterprises keep ISE admin access on dedicated management VLANs or jump hosts
  • ISE is often not directly internet-exposed
  • NGFWs, ACLs, and VPN segmentation commonly narrow the reachable population
Detection/coverage: External attack-surface scanners can find exposed ISE nodes; internal exposure usually needs authenticated asset inventory or config review.
STEP 02

Obtain Read Only Admin credentials

Cisco states exploitation requires at least Read Only Admin credentials. That is the decisive friction point: the attacker is already using privileged administrative identity, whether by phishing, SSO compromise, password reuse, insider abuse, or prior foothold into the admin network.
Conditions required:
  • Valid ISE administrative credentials with Read Only Admin or higher
  • Admin account is not blocked by MFA or source restrictions
Where this breaks in practice:
  • MFA on admin access materially raises attacker cost
  • PAM, jump boxes, IP allowlists, and SSO conditional access often gate admin logins
  • Read-only admin accounts are rarer than end-user accounts
Detection/coverage: Good coverage if admin authentication is centralized: IdP logs, VPN logs, TACACS/RADIUS records, and ISE admin login events should show the access.
STEP 03

Send the crafted HTTP request and pop OS-level execution

With management access and credentials, the attacker can weaponize the bug using a crafted HTTP request from a simple client such as curl, Burp Repeater, or a custom Python/Go script. Cisco attributes the flaw to insufficient input validation; successful exploitation yields user-level access on the underlying OS and then privilege escalation to root.
Conditions required:
  • Authenticated session to the ISE management interface
  • Vulnerable 3.1/3.2/3.3/3.4 release according to Cisco fixed-version guidance
Where this breaks in practice:
  • No public, validated PoC was found in authoritative sources during this review
  • Appliance-specific request formatting may slow opportunistic attackers
  • Inline inspection on management networks is inconsistent across enterprises
Detection/coverage: Cisco lists Snort rules 66300-66306 tied to this advisory. HTTP logs, reverse-proxy logs, and unusual admin API request patterns are the best native detection points.
STEP 04

Abuse root on the NAC control point

Once root is obtained on ISE, the attacker is sitting on a system that brokers network access policy, identity context, and device posture decisions. The technical blast radius can include policy tampering, credential or secret access, and service disruption; in single-node deployments Cisco warns exploitation can make the node unavailable, blocking new endpoint authentications until recovery.
Conditions required:
  • Successful code execution and privilege escalation on the ISE node
  • Target is operationally important in the organization's auth path
Where this breaks in practice:
  • Clustered deployments reduce the availability blast radius versus single-node deployments
  • Post-exploitation dwell time may be shortened by good admin log review and config monitoring
  • Appliance-centric workflows can make persistence less convenient than on general-purpose servers
Detection/coverage: Coverage is weaker here than on standard Linux servers because EDR is often absent on appliances. Watch for unexpected CLI activity, config drift, certificate changes, and ISE service restarts.
03 · Intelligence Metadata

The supporting signals.

In-the-wild statusNo confirmed exploitation found in Cisco's advisory; Cisco PSIRT says it is not aware of public announcements or malicious use as of 2026-04-15.
Proof-of-concept availabilityNo validated public PoC located in authoritative sources reviewed. Some aggregator sites mention exploit tracking or a GitHub advisory, but I did not find a trustworthy public exploit repository to treat as confirmed weaponization.
EPSS0.00469 from the prompt-supplied intel, which is *low* for near-term exploitation likelihood. I did not independently re-pull the exact percentile from FIRST during this review.
KEV statusNot listed in CISA KEV at time of review; no KEV due date applies.
CVSS vector meaningAV:N/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H says easy once you are on the admin plane and authenticated. The problem is that PR:L hides a major real-world fact: here 'low privileges' still means administrative access to ISE, not a throwaway end-user account.
Affected versionsPer Cisco fixed-version guidance: earlier than 3.2, 3.2 before Patch 8, 3.3 before Patch 8, and 3.4 before Patch 4 are affected. Cisco says ISE-PIC is not affected by this CVE.
Fixed versions3.2 Patch 8, 3.3 Patch 8, 3.4 Patch 4; for earlier than 3.2, Cisco says migrate to a fixed release; 3.5 is not vulnerable.
Exposure dataBitsight's public footprint page showed 72 total public-facing Cisco ISE observations over the prior 30 days, including 19 observations on 3.4, 22 on 3.3, and 18 on 3.2. That is not huge internet-scale exposure, but it is enough to prove some orgs still expose ISE externally.
Disclosure timelineCisco advisory first published 2026-04-15 16:00 GMT; the CVE record shows published on 2026-04-15 and NVD still Awaiting Analysis in the reviewed enrichment snapshot.
Reporter / sourceCisco says the issue was found during internal security testing by X.B. of the Cisco Advanced Security Initiatives Group (ASIG), not by an external researcher.
04 · The Call

noisgate verdict.

Final Verdict
DOWNGRADED to HIGH (8.2/10)

The single biggest reason this lands in HIGH instead of CRITICAL is the prerequisite of valid ISE administrative credentials plus management-plane reachability. That sharply narrows the exposed population, but because the target is the enterprise NAC control point and successful exploitation reaches root, the residual risk is still too large for a MEDIUM call.

HIGH Affected/fixed version mapping from Cisco advisory
HIGH Exploit prerequisite of Read Only Admin credentials
MEDIUM Public exploitation / PoC availability assessment

Why this verdict

  • Downgrade from 9.9 because PR:L is misleading in practice: this is not 'any low-priv user'; it is Read Only Admin on ISE, which already implies privileged access to a security appliance.
  • Further downgrade because the attacker needs management-plane reachability: most enterprises do not intentionally expose ISE admin UI to the internet, so reachable population is far smaller than commodity edge-RCE bugs.
  • Kept in HIGH because blast radius is meaningful once landed: root on ISE means compromise of a central identity and network-access enforcement point, and Cisco warns single-node deployments can also suffer auth-disrupting DoS.
  • No upgrade pressure from threat intel: no KEV listing, low prompt-supplied EPSS, and no authoritative evidence of active exploitation found during review.
  • No bigger downgrade because compensating stack is imperfect on appliances: EDR is often absent, logging can be thinner than on Windows/Linux servers, and admin credential theft remains a realistic path.

Why not higher?

This is not pre-auth, not broadly wormable, and not something a random internet scanner can usually turn into instant root without prior progress. Requiring both admin-plane access and Read Only Admin credentials makes it a compound chain, and each prerequisite knocks down the real-world attack volume.

Why not lower?

Once the prerequisites are met, exploitation is low-complexity and the payoff is root on the NAC brain. Even though this is a narrower population than classic edge RCE, the operational blast radius is still serious enough that calling it MEDIUM would understate the risk to enterprises that actually run Cisco ISE.

05 · Compensating Control

What to do — in priority order.

  1. Lock admin access to jump hosts — Restrict ISE web management reachability to dedicated admin subnets, bastions, or VPN source ranges only. This is the cleanest exposure reduction while you schedule patching, and for a HIGH verdict it should be in place within 30 days if not already standard.
  2. Enforce MFA on every ISE admin path — This bug starts with valid admin credentials, so make credential theft harder to convert into appliance compromise. If SSO or VPN gates ISE administration, require phishing-resistant MFA where possible and close gaps within 30 days.
  3. Cull read-only admin accounts — Treat Read Only Admin as code-exec-capable for this advisory, not as a harmless audit role. Remove stale accounts, rotate secrets for service-linked admins, and shrink group membership within 30 days.
  4. Deploy Cisco detection content — Enable and monitor the Cisco-published Snort rules 66300-66306 and alert on unusual admin HTTP requests to ISE. Detection is not remediation, but it gives you a chance to catch exploit attempts during the 30-day mitigation window.
  5. Prioritize single-node deployments — Cisco explicitly notes single-node ISE deployments can become unavailable and block new endpoint authentications after exploitation. Identify those appliances first because their business impact is worse even though the exploit preconditions are the same.
What doesn't work
  • Read Only Admin role separation does not help here, because Cisco says that exact role is sufficient to exploit the bug.
  • A generic perimeter WAF is not a dependable answer when the management UI is internal, behind VPN, or not actually traversing the WAF path.
  • Credential rotation without reachability reduction leaves the same attack path available to the next stolen admin account.
  • Vuln scanning alone is not mitigation; it tells you where ISE is, but it does nothing to block a valid admin session from sending the crafted request.
06 · Verification

Crowdsourced verification payload.

Run this from an auditor workstation or jump host that can SSH to the Cisco ISE CLI, not from a random endpoint. Invoke it as ./check_cve_2026_20180.sh [email protected]; it needs only enough CLI privilege to run show version, not root shell access.

noisgate-verify.sh
BASHREAD-ONLYSAFE
#!/usr/bin/env bash
# check_cve_2026_20180.sh
# Determine likely exposure to CVE-2026-20180 on a Cisco ISE node via SSH CLI.
# Exit codes: 0=PATCHED, 1=VULNERABLE, 2=UNKNOWN, 3=USAGE/CONNECT ERROR

set -u

TARGET="${1:-}"
if [[ -z "$TARGET" ]]; then
  echo "Usage: $0 <user@host>" >&2
  exit 3
fi

SSH_OPTS=(
  -o BatchMode=yes
  -o ConnectTimeout=10
  -o StrictHostKeyChecking=accept-new
)

get_version_output() {
  ssh "${SSH_OPTS[@]}" "$TARGET" "show version" 2>/dev/null
}

OUT="$(get_version_output)"
if [[ -z "$OUT" ]]; then
  echo "UNKNOWN - unable to retrieve 'show version' from $TARGET"
  exit 2
fi

# Normalize whitespace for easier regex handling.
NORM="$(printf '%s\n' "$OUT" | tr -s '[:space:]' ' ')"

# Try to find a major.minor release like 3.1 / 3.2 / 3.3 / 3.4 / 3.5
REL=""
PATCH="0"

if [[ "$NORM" =~ ([0-9]+\.[0-9]+) ]]; then
  REL="${BASH_REMATCH[1]}"
fi

if [[ "$NORM" =~ [Pp]atch[[:space:]]*([0-9]+) ]]; then
  PATCH="${BASH_REMATCH[1]}"
fi

if [[ -z "$REL" ]]; then
  echo "UNKNOWN - could not parse version from output"
  printf '%s\n' "$OUT"
  exit 2
fi

status="UNKNOWN"
reason="unhandled version logic"

case "$REL" in
  3.5)
    status="PATCHED"
    reason="Cisco advisory states 3.5 is not vulnerable"
    ;;
  3.4)
    if (( PATCH >= 4 )); then
      status="PATCHED"
      reason="3.4 Patch 4 or later is fixed"
    else
      status="VULNERABLE"
      reason="3.4 before Patch 4 is vulnerable"
    fi
    ;;
  3.3)
    if (( PATCH >= 8 )); then
      status="PATCHED"
      reason="3.3 Patch 8 or later is fixed"
    else
      status="VULNERABLE"
      reason="3.3 before Patch 8 is vulnerable"
    fi
    ;;
  3.2)
    if (( PATCH >= 8 )); then
      status="PATCHED"
      reason="3.2 Patch 8 or later is fixed"
    else
      status="VULNERABLE"
      reason="3.2 before Patch 8 is vulnerable"
    fi
    ;;
  3.1)
    status="VULNERABLE"
    reason="Earlier than 3.2 must migrate to a fixed release"
    ;;
  *)
    # Earlier than 3.2 or any unexpected branch
    if [[ "$REL" =~ ^([0-2])\. ]]; then
      status="VULNERABLE"
      reason="Earlier than 3.2 must migrate to a fixed release"
    else
      status="UNKNOWN"
      reason="Version outside scripted Cisco guidance"
    fi
    ;;
esac

echo "$status - $TARGET reports release $REL patch $PATCH ($reason)"

case "$status" in
  PATCHED) exit 0 ;;
  VULNERABLE) exit 1 ;;
  *) exit 2 ;;
esac
07 · Bottom Line

If you remember one thing.

TL;DR
Monday morning, treat this as a credentialed management-plane RCE on a crown-jewel appliance: identify every ISE node, flag any single-node deployments, and immediately verify whether the admin UI is reachable from anything broader than your jump hosts. For a HIGH verdict the noisgate mitigation SLA is ≤30 days, so within that window lock administration to trusted source ranges, enforce MFA, and reduce or remove read-only admin accounts; the noisgate remediation SLA is ≤180 days, and that means moving to 3.2 Patch 8, 3.3 Patch 8, 3.4 Patch 4, or 3.5 well before the long tail of appliance patching pushes this into backlog purgatory.

Sources

  1. Cisco advisory: Cisco Identity Services Engine Remote Code Execution Vulnerabilities
  2. OpenCVE enrichment for CVE-2026-20180
  3. CISA Known Exploited Vulnerabilities Catalog
  4. Bitsight Groma Explorer: Cisco Identity Services Engine Observation Footprint
  5. SecurityWeek coverage of Cisco April 2026 patches
  6. CIRCL Vulnerability Lookup for CVE-2026-20180
  7. ZeroPath summary of CVE-2026-20180
Peer Review

What defenders are saying.

Submit a review attribution: handle + country only
0 flags selected · stored anonymously
Validation Results

Crowdsourced verification outputs.

Results submitted by users who ran the verification payload against their environment.