Skip to content
CVE-2026-17134

IBM API Connect SQL Injection (CVE-2026-17134)

Initial Access Collection Last updated:

Detects exploitation attempts against CVE-2026-17134, a SQL injection vulnerability (CWE-89) in IBM API Connect versions 12.1.0.0 through 12.1.1.2. A remote authenticated attacker can inject arbitrary SQL commands via vulnerable API parameters, potentially reading or modifying backend database contents. This detection surfaces SQL-injection patterns in HTTP requests to API Connect management/portal endpoints by authenticated users, anomalous SQL error responses, and database-side signs of injected queries. CVSS 9.9; fixed in 12.1.1.3.

Vulnerability Intelligence

Theoretical

Affected Software

Vendor
ibm
Product
api_connect
Versions
>= 12.1.0.0, < 12.1.1.3

Weakness (CWE)

Timeline

Disclosed
November 7, 2026

References & Proof of Concept

CVSS

9.9
Critical (9.0–10)

What is CVE-2026-17134 IBM API Connect SQL Injection (CVE-2026-17134)?

IBM API Connect SQL Injection (CVE-2026-17134) (CVE-2026-17134) maps to the Initial Access and Collection tactics — the adversary is trying to get into your network in MITRE ATT&CK.

This page provides production-ready detection logic for IBM API Connect SQL Injection (CVE-2026-17134), covering the data sources and telemetry it touches: IIS Logs, Web Server Logs, Microsoft Defender for Endpoint. The queries below are rated critical severity at medium confidence, and ship for 7 SIEM platforms — KQL, SPL, Elastic, QRadar, Sumo, YARA-L, LogScale.

MITRE ATT&CK

Tactic
Initial Access Collection
Microsoft Sentinel / Defender
kusto
// df00tech DF00-fd437e8466 · https://df00tech.com/detections/CVE-2026-17134 · licence: https://df00tech.com/detections/licence
let sqliPatterns = dynamic(["union select","' or '1'='1","'--","' or 1=1","waitfor delay","pg_sleep","sleep(","benchmark(","information_schema","';--","/*!","xp_cmdshell","convert(int","cast(","' and '1'='1"]);
W3CIISLog
| where csUriStem has_any ("/api/", "/consumer-api/", "/cloud/", "/portal/")
| extend decodedQuery = tolower(url_decode(csUriQuery))
| extend decodedUri = tolower(url_decode(csUriStem))
| where decodedQuery has_any (sqliPatterns) or decodedUri has_any (sqliPatterns)
| where isnotempty(csUsername) and csUsername != "-"
| project TimeGenerated, cIP, csUsername, csMethod, csUriStem, csUriQuery, scStatus, sSiteName, Computer
| sort by TimeGenerated desc

Flags authenticated HTTP requests to IBM API Connect endpoints whose URI or query string contains SQL-injection syntax after URL decoding.

critical severity medium confidence

Data Sources

IIS Logs Web Server Logs Microsoft Defender for Endpoint

Required Tables

W3CIISLog

False Positives

  • Developers or QA running security test suites (DAST scanners) against staging API Connect instances.
  • Legitimate API payloads that contain SQL-like keywords (e.g., analytics queries submitted as parameters).
  • Automated vulnerability scanners authorized by the security team.

Query fingerprint DF00-fd437e8466 · licence

Sigma logsource & community rules

The detection logic for IBM API Connect SQL Injection (CVE-2026-17134) (CVE-2026-17134) above is provided in a vendor-neutral form so you can deploy it on any SIEM. The same logic is shipped here as native KQL (Microsoft Sentinel / Defender), SPL (Splunk), Elastic (Elastic Security (EQL)), QRadar (IBM QRadar (AQL)), Sumo (Sumo Logic CSE), YARA-L (Google Chronicle / SecOps), LogScale (CrowdStrike LogScale (CQL)) queries. In Sigma terms, this detection targets the following logsource:

logsource:
  product: windows

Browse the community-maintained Sigma rules for this technique:


Testing Methodology

Validate this detection against 3 adversary techniques from Atomic Red Team. Each test below lists the behaviour to exercise and the telemetry you should expect to see. Executable commands and cleanup steps are further down this page.

  1. Test 1 Error-based SQL injection probe against API Connect consumer API

    Expected signal: Web/IIS access log entry with authenticated user, URI /consumer-api/orgs, query string containing url-encoded "' OR '1'='1'--", and likely HTTP 400/500 response.

  2. Test 2 Union-based SQL injection probe

    Expected signal: Access log entry with decoded query containing 'union select' and 'information_schema' for an authenticated user.

  3. Test 3 Time-based blind SQL injection probe

    Expected signal: Access log entry with decoded query containing 'pg_sleep' and ';' for an authenticated user, with elevated response time.


Response Playbook

Triage

  1. Confirm the target host runs IBM API Connect at a vulnerable version (12.1.0.0 through 12.1.1.2); check build/version via the admin console or /cloud/api/cloud/settings.
  2. Identify the authenticated user account tied to the flagged request and determine whether it is a legitimate consumer/provider org member or a compromised/over-privileged account.
  3. URL-decode the captured query string and manually assess whether the payload is genuine SQL-injection syntax versus a benign keyword match.
  4. Correlate the source IP and user agent against known scanner/DAST infrastructure and the user's normal access baseline.

Containment

  1. Suspend or disable the implicated API Connect user account and revoke its active OAuth/API tokens.
  2. Block the attacker source IP at the WAF/load balancer fronting the API Connect management and portal endpoints.
  3. If exploitation is confirmed, isolate the API Connect management subsystem and its backing database from untrusted network segments pending patch to 12.1.1.3.

Evidence Collection

  1. Preserve IIS/web/ingress access logs showing the full request URI, query string, authenticated user, source IP, and response codes.
  2. Capture API Connect management/portal application logs and the backing PostgreSQL/database query and error logs covering the time window.
  3. Snapshot the API Connect database for forensic comparison to detect unauthorized reads or schema/data modification.

Escalation Criteria

  • ! Escalate to incident response if SQL error responses (HTTP 500 with DB stack traces) or time-based injection latency confirm the injection executed.
  • ! Escalate if evidence shows data exfiltration, credential/token theft, or modification of consumer/provider org records.
  • ! Escalate to the platform owner if the account used was privileged (org admin) or if multiple accounts/IPs show coordinated injection attempts.

Investigation Guide

Forensic Artifacts

  • > Web/IIS access logs containing the SQLi request URI and query string
  • > API Connect application and database (PostgreSQL) query/error logs
  • > Authentication logs for the implicated user account and token issuance records

Tuning Guidance

Baseline legitimate API Connect consumers that submit SQL-like keywords in parameters and allowlist known DAST/scanner source IPs used by the security team. Tighten the endpoint list to the actual API Connect management/portal paths in your deployment, and raise confidence to high when SQLi syntax coincides with HTTP 500 database-error responses or measurable time-based latency. Prefer URL-decoded matching to avoid evasion via encoding and to reduce raw-keyword noise.


Hunting Queries

Surfaces accounts/IPs making repeated SQL-injection-style requests to API Connect, indicating sustained probing rather than a single false positive.

Hunting — KQL
kql
// df00tech DF00-fd437e8466 · https://df00tech.com/detections/CVE-2026-17134 · licence: https://df00tech.com/detections/licence
W3CIISLog | where csUriStem has_any ("/api/","/consumer-api/","/portal/") | extend q=tolower(url_decode(csUriQuery)) | where q has_any ("union select","information_schema","pg_sleep","waitfor delay") | summarize count(), make_set(csUriStem) by csUsername, cIP | where count_ > 3
Hunting — SPL
spl
``` df00tech DF00-fd437e8466 · https://df00tech.com/detections/CVE-2026-17134 · licence: https://df00tech.com/detections/licence ```
index=web (uri_path="*/api/*" OR uri_path="*/portal/*") | eval q=urldecode(lower(uri_query)) | where match(q,"union\s+select|information_schema|pg_sleep|waitfor\s+delay") | stats count values(uri_path) by user, src_ip | where count>3

Atomic Red Team Tests

Test 1 Error-based SQL injection probe against API Connect consumer API
linux

Sends an authenticated request with a single-quote-based error payload to an API Connect endpoint to trigger a database error, simulating CVE-2026-17134 exploitation.

Command

bash
curl -s -k -H "Authorization: Bearer $APIC_TOKEN" "https://apic.lab.local/consumer-api/orgs?filter=name%20eq%20%27%27%20OR%20%271%27%3D%271%27--%20"

Cleanup

bash
echo 'No persistent changes; clear test token: unset APIC_TOKEN'

Expected Telemetry

Web/IIS access log entry with authenticated user, URI /consumer-api/orgs, query string containing url-encoded "' OR '1'='1'--", and likely HTTP 400/500 response.

Expected Detection

KQL/SPL/EQL rules match the decoded 'or '1'='1 and '-- patterns on an authenticated API Connect request.

Test 2 Union-based SQL injection probe
linux

Attempts a UNION SELECT injection via a query parameter to extract additional columns, modeling data-exfiltration exploitation of the SQLi flaw.

Command

bash
curl -s -k -H "Authorization: Bearer $APIC_TOKEN" "https://apic.lab.local/api/catalogs?q=1%20UNION%20SELECT%20username,password%20FROM%20information_schema.tables--"

Cleanup

bash
echo 'No persistent changes to clean up'

Expected Telemetry

Access log entry with decoded query containing 'union select' and 'information_schema' for an authenticated user.

Expected Detection

All seven SIEM queries match the 'union select' and 'information_schema' patterns.

Test 3 Time-based blind SQL injection probe
linux

Sends a time-delay payload (pg_sleep) to infer injectability through response latency, simulating blind exploitation.

Command

bash
curl -s -k -o /dev/null -w "%{time_total}\n" -H "Authorization: Bearer $APIC_TOKEN" "https://apic.lab.local/portal/search?term=x%27%3BSELECT%20pg_sleep(5)--%20"

Cleanup

bash
echo 'No persistent changes; no cleanup required'

Expected Telemetry

Access log entry with decoded query containing 'pg_sleep' and ';' for an authenticated user, with elevated response time.

Expected Detection

Queries match the 'pg_sleep' time-based injection pattern on an authenticated API Connect request.

Related Detections

Browse by tactic

Full ATT&CK matrix Search all detections