Leaving the master key under the doormat, but only tenants who already live in the building can grab it
CVE-2026-67100 bundles two flaws in HCL BigFix Service Management V23: a classic SQL injection (CWE-89) that lets an authenticated user run arbitrary database commands and extract sensitive system internals, and a cross-tenant data exposure flaw (CWE-200/CWE-284) that lets one tenant's user manipulate request identifiers to pull full PII and personal profile data belonging to *other* organizations hosted on the same instance. Together, these turn any low-privilege seat into a data-exfiltration vector across tenant boundaries.
HCL scored this a 9.8 CRITICAL using CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H, but the advisory's own description states the attacker must be *authenticated*. That makes PR:N a misclassification — the correct privileges-required value is at least PR:L, which by itself drops the CVSS to ~8.8. Factor in the narrow installed base of BigFix SM (a niche ITSM competitor to ServiceNow and Jira SM), the absence of any public proof-of-concept, zero KEV listing, and no confirmed in-the-wild exploitation, and the effective risk for most enterprises falls to HIGH, not CRITICAL. That said, for MSPs and shared-hosting operators running multi-tenant BigFix SM, the cross-tenant PII leak is a genuinely dangerous finding that warrants urgent action.
4 steps from start to impact.
Obtain valid BigFix SM credentials
- Valid BigFix SM user account on the target instance
- Network access to the BigFix SM web interface (typically HTTPS/443)
- BigFix SM portals are rarely exposed to the public internet — most deployments sit behind VPN or corporate network boundaries
- MFA on the SM login portal blocks credential-spray and phished-password scenarios
- The product's relatively small market share means fewer attacker tools and playbooks target it specifically
Identify injectable parameter
- Authenticated session from Step 1
- Knowledge of or ability to fuzz SM API/form endpoints
- No public exploit code or Nuclei template exists as of disclosure date
- Web Application Firewalls (WAFs) with SQL injection rule sets will catch many common injection patterns
- Without a known parameter, discovery requires non-trivial effort
', UNION, SELECT, --); database slow-query or error logs showing malformed queriesExtract sensitive data via SQL injection
UNION SELECT or error-based/blind injection payloads to enumerate database schemas, extract credentials, configuration secrets, and tenant data. Because BigFix SM stores ITSM ticket data, asset records, and user PII, the database is a rich target. The attacker can also target internal system tables to harvest database credentials or escalate within the DB tier.- Confirmed injectable parameter from Step 2
- Database user context used by the application has read access to sensitive tables
- Database-level least-privilege (if the app DB user is read-only on its own schema) limits lateral movement within the DB
- Parameterized query enforcement at the ORM layer may protect some endpoints even if others are vulnerable
INFORMATION_SCHEMA queries or bulk SELECT across tenant tables; DLP alerts on large data exports from SMCross-tenant PII exfiltration
- Authenticated session
- Multi-tenant deployment (MSP or shared-hosting model)
- Single-tenant deployments are not affected by the cross-tenant component
- Organizations using dedicated BigFix SM instances per business unit eliminate this vector entirely
The supporting signals.
| In-the-Wild Exploitation | No confirmed exploitation as of 2026-09-18. Not listed in CISA KEV. No threat-intelligence vendor (Mandiant, CrowdStrike, Unit 42) has reported active campaigns. |
|---|---|
| Proof-of-Concept | None public. No GitHub repos, no Nuclei templates, no Metasploit modules. The SQLi is described generically — no specific parameter or endpoint has been disclosed. |
| EPSS Score | Not yet scored. CVE was published 2026-09-18 (today). EPSS typically populates within 24–72 hours of NVD ingestion. |
| KEV Status | Not listed in CISA KEV catalog as of 2026-09-18. |
| CVSS Vector | CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H — 9.8 CRITICAL per vendor. However, the advisory states *authenticated attacker*, which contradicts PR:N. Corrected to PR:L the score drops to ~8.8. |
| Affected Versions | HCL BigFix Service Management V23 (all builds prior to the fix referenced in KB0133782). Specific sub-version ranges not disclosed by HCL. |
| Fixed Version | Patch referenced in HCL KB0133782, published 2026-09-18. Exact version string not publicly enumerated; check the KB article with your HCL support entitlement. |
| Internet Exposure | Minimal. BigFix Service Management is an internal ITSM portal. Shodan/Censys/FOFA show negligible public-facing instances. The vast majority of deployments sit behind VPN or corporate firewalls. |
| Disclosure Date | 2026-09-18 (coordinated disclosure by HCL Software). |
| Reporter | Not publicly attributed. HCL's security bulletin does not credit an external researcher. |
noisgate verdict.
The single most decisive factor is the CVSS vector misclassification: the vendor scored PR:N (no privileges required) while the advisory's own text explicitly requires an *authenticated* attacker, which compoundingly narrows the reachable attack surface. This authentication prerequisite, combined with a niche installed base and zero exploit availability, pulls the effective severity from CRITICAL to HIGH.
Why this verdict
- CVSS vector mismatch (PR:N → PR:L): The advisory explicitly states 'authenticated attacker,' yet the CVSS vector claims no privileges required. Correcting to
PR:Lalone drops the base score from 9.8 to ~8.8. This is the single largest scoring error. - Narrow installed base: BigFix Service Management is a niche ITSM product competing with ServiceNow, Jira SM, and BMC Helix. It manages 155M+ endpoints as a *platform* family, but the Service Management module has a small fraction of that footprint. Fewer targets means fewer real-world attack opportunities.
- No exploit availability: Zero public PoCs, no Metasploit/Nuclei modules, no named researchers. Weaponization requires original vulnerability research against a relatively obscure codebase, raising the practical bar significantly.
- Role multiplier: BigFix *Service Management* is a ticketing and workflow tool — a line-of-business application (role b). It is not the BigFix endpoint-management agent/server that controls patching across fleets. Compromise of SM yields ITSM ticket data and user PII, not remote code execution on managed endpoints. The blast radius is *tenant-level data exposure*, not fleet compromise. For MSPs running multi-tenant instances, the cross-tenant PII leak elevates risk within that deployment model, but does not reach domain-takeover or supply-chain-pivot scale. This keeps the floor at HIGH, not CRITICAL.
- Internet exposure near zero: BigFix SM is deployed as an internal portal. Scanner data (Shodan/Censys) shows negligible public exposure, meaning attackers must already have internal network access or VPN credentials before reaching the login page — an additional friction layer not captured in the CVSS vector.
Why not higher?
The vulnerability cannot remain CRITICAL because the CVSS vector's PR:N is contradicted by the advisory's own requirement for authentication, no public exploit exists, the product has a small installed base, and the blast radius is limited to data within the ITSM platform rather than fleet-scale code execution. Even for multi-tenant MSP deployments where cross-tenant PII exposure is serious, the chain does not escalate to domain takeover, supply-chain pivot, or infrastructure compromise — it stays at data exposure scope.
Why not lower?
SQL injection with cross-tenant data exposure in a product that stores PII and ITSM records across organizations is a genuine privacy and compliance hazard. The authentication barrier lowers the bar but does not eliminate it — any employee with an SM account is a potential insider threat vector. The SQLi component could yield database credentials or configuration secrets that enable lateral movement beyond the SM tier. For MSPs, the cross-tenant angle makes this a contractual and regulatory emergency (GDPR, SOC 2 tenant-isolation controls). These factors establish a firm HIGH floor.
What to do — in priority order.
- Restrict BigFix SM access to VPN/internal network only — If SM is reachable from the internet (even behind a CDN or reverse proxy), pull it behind VPN immediately. This eliminates unauthenticated network reachability and raises the cost of exploitation to post-initial-access. Deploy within the noisgate HIGH mitigation SLA of 30 days.
- Deploy WAF rules blocking SQL injection patterns on SM endpoints — Enable OWASP CRS or equivalent SQLi detection rules on any WAF or reverse proxy fronting BigFix SM. This catches the most common injection payloads (
UNION SELECT,' OR 1=1, stacked queries). Deploy within 30 days per the noisgate mitigation SLA. - Enforce MFA on all BigFix SM accounts — Since the attack requires authentication, adding MFA to the SM login eliminates credential-spray and phished-password paths. This is the single highest-ROI compensating control for this CVE. Deploy within 30 days.
- Audit and restrict database user privileges for the SM application — Ensure the database service account used by BigFix SM has minimal privileges — read/write only on its own schema, no access to
INFORMATION_SCHEMAor cross-database queries. This limits SQLi blast radius even if injection succeeds. - Enable database audit logging and alert on cross-tenant queries — Turn on query-level audit logging on the SM backend database. Alert on queries that reference tenant identifiers outside the authenticated user's scope, or that touch system catalog tables.
- Network segmentation between SM and BigFix Platform servers — while good hygiene, it does not prevent SQLi or cross-tenant data access within SM's own database. The vulnerability is application-layer, not network-layer.
- Endpoint Detection and Response (EDR) — EDR agents on the SM server won't detect SQL injection; it's valid HTTP traffic that abuses application logic. EDR only helps if the attacker pivots to OS-level post-exploitation after extracting DB credentials.
- Patching the BigFix Platform agent/server — BigFix Service Management is a separate product from BigFix Platform. Patching the endpoint-management stack does not remediate the SM vulnerability.
Crowdsourced verification payload.
Run this script on the BigFix Service Management application server (or any host with curl and network access to SM). Invoke as: sudo bash check_cve_2026_67100.sh https://bigfix-sm.corp.local. Requires curl and read access to the SM installation directory if checking locally. No authentication needed for the version-file check path.
#!/usr/bin/env bash
# check_cve_2026_67100.sh — Detect CVE-2026-67100 in HCL BigFix Service Management
# Usage: bash check_cve_2026_67100.sh [SM_BASE_URL]
# Exit codes: 0=PATCHED, 1=VULNERABLE, 2=UNKNOWN
set -euo pipefail
SM_URL="${1:-}"
RESULT="UNKNOWN"
# Method 1: Check version via SM API/about endpoint
if [[ -n "$SM_URL" ]]; then
echo "[*] Checking BigFix SM version at $SM_URL ..."
VERSION=$(curl -sk --max-time 10 "${SM_URL}/api/v1/version" 2>/dev/null \
| grep -oP '"version"\s*:\s*"\K[^"]+' || true)
if [[ -z "$VERSION" ]]; then
VERSION=$(curl -sk --max-time 10 "${SM_URL}/about" 2>/dev/null \
| grep -ioP 'version[:\s]+\K[0-9]+\.[0-9]+\.[0-9]+[^\s<"]*' | head -1 || true)
fi
if [[ -n "$VERSION" ]]; then
echo "[*] Detected version: $VERSION"
# HCL has not published exact fixed sub-version; flag all V23.x as potentially vulnerable
if echo "$VERSION" | grep -qP '^23\.'; then
echo "[!] Version $VERSION is in the affected V23 range."
echo "[!] Check HCL KB0133782 to confirm if your exact build includes the fix."
RESULT="VULNERABLE"
else
echo "[+] Version $VERSION does not appear to be in the affected V23 range."
RESULT="PATCHED"
fi
else
echo "[?] Could not retrieve version from $SM_URL. Trying local file check..."
fi
fi
# Method 2: Local file version check (run on SM app server)
if [[ "$RESULT" == "UNKNOWN" ]]; then
INSTALL_DIRS=("/opt/hcl/bigfix-sm" "/opt/BigFix/SM" "/usr/local/bigfix-sm")
for DIR in "${INSTALL_DIRS[@]}"; do
if [[ -d "$DIR" ]]; then
echo "[*] Found installation at $DIR"
VERFILE=$(find "$DIR" -maxdepth 3 -name 'version*.properties' -o -name 'build.info' 2>/dev/null | head -1)
if [[ -n "$VERFILE" ]]; then
echo "[*] Version file: $VERFILE"
cat "$VERFILE"
if grep -qiP '23\.' "$VERFILE"; then
RESULT="VULNERABLE"
echo "[!] VULNERABLE — V23 detected. Verify patch status against KB0133782."
else
RESULT="PATCHED"
echo "[+] PATCHED — version outside affected range."
fi
fi
break
fi
done
fi
if [[ "$RESULT" == "UNKNOWN" ]]; then
echo "[?] UNKNOWN — Could not determine BigFix SM version."
echo " Manually verify against HCL KB0133782."
exit 2
elif [[ "$RESULT" == "VULNERABLE" ]]; then
echo "VULNERABLE"
exit 1
else
echo "PATCHED"
exit 0
fiIf you remember one thing.
Sources
What defenders are saying.
Crowdsourced verification outputs.
Results submitted by users who ran the verification payload against their environment.