CVE-2026-96429

Executive Summary

CVE-2026-96429 is a SQL injection vulnerability in the /WebAgenda/SMBAjaxConfigProcess.do API endpoint of Flowring Agentflow 4.0, exploitable remotely via the id parameter. Any organization exposing the affected Agentflow 4.0 build (prior to the 2025/08/08 release) to untrusted networks is at risk of unauthenticated arbitrary SQL command execution. Successful exploitation can lead to full database compromise, data exfiltration, and potential downstream lateral movement within the hosting environment.

Technical Analysis

The vulnerability resides in the SMBAjaxConfigProcess.do endpoint exposed under the /WebAgenda/ web path. The NVD description confirms that the id parameter is passed to a backend SQL query without adequate sanitization or parameterization, resulting in a classic CWE-89 SQL Injection condition. Because the endpoint is reachable over the network and the injection point is exposed before any effective authentication barrier (no privileges are noted as required), an attacker can supply crafted input through the id parameter to alter the semantics of the underlying SQL statement.

Flowring Agentflow is a workflow/business process automation platform commonly deployed on internal enterprise networks for form routing, document management, and audit reporting. The WebAgenda module appears to be one of its web-facing components, and SMBAjaxConfigProcess implies an AJAX-style configuration handler that processes requests server-side. The .do extension is characteristic of Java servlet/Struts-style controllers, though the CNA did not provide implementation details, so the exact ORM, driver, or database (e.g., MSSQL, Oracle, MySQL) is not specified in available data.

The criticality stems from three properties of the flaw:

1. Unauthenticated reachability — no prior credential or session required per the description.
2. Arbitrary SQL execution — not limited to a specific column or boolean condition.
3. Remote exploitability — an attacker only needs HTTP access to the endpoint.

No public proof-of-concept, Nuclei template, or exploit code is currently available, so precise payload construction must be inferred from the vulnerability class. Standard SQLi classes applicable here include UNION-based extraction, boolean/time-based blind injection, and — depending on the backend DBMS and driver configuration — stacked queries leading to command execution via xp_cmdshell (MSSQL) or equivalent features.

Exploitation Scenarios

No public PoC is available; the attack chain below is inferred strictly from the vulnerability class and NVD description.

1. Reconnaissance / Target Identification An attacker scans for internet- or intranet-facing Agentflow deployments. Fingerprinting typically relies on distinctive HTTP response content (Agentflow/WebAgenda branding), the SMBAjaxConfigProcess.do path, or Shodan/Censys queries such as http.html:"WebAgenda". Because there is no CVE-specific signature yet, discovery is opportunistic against known Flowring deployments.

2. Triggering the Vulnerability The attacker sends an HTTP request to /WebAgenda/SMBAjaxConfigProcess.do with a crafted id parameter. No authentication is described as required, so the request can come from any network-reachable position.

3. Payload Execution (vulnerability-class level) The attacker manipulates the id value to alter the SQL statement executed server-side. Techniques applicable to arbitrary SQLi include:

• Boolean-based blind: inject tautologies/contradictions to infer data bit-by-bit.
• UNION-based: append a UNION SELECT to return attacker-chosen columns directly in the response.
• Time-based blind: inject delays to confirm injection points when responses are opaque.
• Stacked queries (DBMS-dependent): execute multiple statements, potentially leading to OS-level command execution via database features.

4. Post-Exploitation Impact Successful exploitation grants the attacker read and/or write access to whatever the application's database account is permitted to touch. Typical outcomes include credential/configuration theft, modification of workflow or audit data, and — via DBMS features or chained primitives — host-level compromise. Given Agentflow is a process automation platform, tampering with workflow data can have integrity and compliance consequences beyond simple data theft.

Impact Assessment

• Confidentiality (High): Arbitrary SQL commands can return data from any table readable by the application's DB account.
• Integrity (High): Attacker can modify or delete records, inject users, or alter workflow state.
• Availability (High): Destructive SQL (DROP/DELETE/UPDATE) or resource-exhausting queries can render the platform unusable.
• Scope (Unchanged): Impact is confined to the Agentflow application's security authority in the CVSS interpretation, though database-level features may extend blast radius in practice.

Blast radius: Organizations running Flowring Agentflow 4.0 (pre-2025/08/08) with the WebAgenda module exposed. Because Agentflow typically handles business-critical workflows (approvals, document routing, audit trails), compromise has direct operational, financial, and regulatory impact. Any single unauthenticated attacker with HTTP reach can exploit the flaw; no user interaction or credentials are required.

CVSS score is listed as N/A in the source data; the vector above reflects the standard unauthenticated network SQLi profile (AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H) but should be treated as an analytical estimate rather than an official score.

Detection Guidance

Detection should focus on anomalous requests to /WebAgenda/SMBAjaxConfigProcess.do — particularly suspicious characters in the id parameter, error responses indicative of SQL parsing failures, and unusual outbound database activity from the web tier.

Sigma Rule

title: SQL Injection Attempt Against Flowring Agentflow SMBAjaxConfigProcess.do
id: 7f3b2a94-9642-4c1e-8a51-agentflow-sqli
status: experimental
description: Detects SQL injection attempts against the Flowring Agentflow WebAgenda SMBAjaxConfigProcess.do endpoint via the id parameter.
references:
  - https://nvd.nist.gov/vuln/detail/CVE-2026-96429
author: Threat Intelligence
date: 2026/01/01
logsource:
  category: webserver
detection:
  selection_endpoint:
    cs-uri-stem|endswith: '/WebAgenda/SMBAjaxConfigProcess.do'
  selection_payload:
    cs-uri-query|contains:
      - "'"
      - '"'
      - '--'
      - 'UNION'
      - 'SELECT'
      - 'OR 1=1'
      - 'SLEEP('
      - 'WAITFOR DELAY'
      - 'xp_cmdshell'
  condition: selection_endpoint and selection_payload
fields:
  - c-ip
  - cs-uri-stem
  - cs-uri-query
  - sc-status
falsepositives:
  - Legitimate application testing
  - Vulnerability scanner traffic (authorized)
level: high

YARA Rule

rule CVE_2026_96429_Flowring_Agentflow_SQLi
{
    meta:
        description = "Detects SQL injection attempts targeting Flowring Agentflow SMBAjaxConfigProcess.do id parameter"
        reference = "https://nvd.nist.gov/vuln/detail/CVE-2026-96429"
        author = "Threat Intelligence"
        date = "2026-01-01"
        severity = "high"

    strings:
        $endpoint = "/WebAgenda/SMBAjaxConfigProcess.do" ascii wide nocase
        $p1 = "UNION SELECT" ascii wide nocase
        $p2 = "OR 1=1" ascii wide nocase
        $p3 = "WAITFOR DELAY" ascii wide nocase
        $p4 = "xp_cmdshell" ascii wide nocase
        $p5 = "' OR '" ascii wide nocase
        $p6 = "SLEEP(" ascii wide nocase

    condition:
        $endpoint and any of ($p*)
}

Suricata IDS Rule

alert http any any -> any any (msg:"CVE-2026-96429 Flowring Agentflow SQLi Attempt SMBAjaxConfigProcess.do"; flow:established,to_server; http.uri; content:"/WebAgenda/SMBAjaxConfigProcess.do"; http.uri; content:"id="; http.uri; pcre:"/(id=[^&]*(%27|'|--|UNION|SELECT|OR\+1%3D1|WAITFOR|SLEEP\())/i"; classtype:web-application-attack; reference:cve,2026-96429; sid:202696429; rev:1;)

Elastic Detection

sequence by source.ip with maxspan=30s
  [network where url.path == "/WebAgenda/SMBAjaxConfigProcess.do"
     and url.query rlike "(?i)(id=.*(%27|'|--|UNION|SELECT|WAITFOR|SLEEP\())"]

Splunk SPL Query

index=web OR index=proxy
| eval uri = coalesce(url, cs_uri_stem, uri_path)
| eval query = coalesce(url_query, cs_uri_query, uri_query)
| where like(uri, "%/WebAgenda/SMBAjaxConfigProcess.do%")
| where match(query, "(?i)(id=.*(%27|'|--|UNION[+ ]SELECT|WAITFOR[+ ]DELAY|SLEEP\\(|OR[+ ]1%3D1|OR[+ ]1=1))")
| stats count min(_time) as first_seen max(_time) as last_seen values(src_ip) as sources values(query) as payloads by uri
| where count > 0
| sort - count

Wazuh OSSEC Rule

<group name="web,appsec,sqli,">
  <rule id="100964" level="12">
    <if_sid>31100</if_sid>
    <url>/WebAgenda/SMBAjaxConfigProcess.do</url>
    <match>id=</match>
    <regex type="pcre2">(?i)(%27|'|--|UNION|SELECT|WAITFOR|SLEEP\(|OR\+1%3D1)</regex>
    <description>CVE-2026-96429: SQL injection attempt against Flowring Agentflow SMBAjaxConfigProcess.do id parameter</description>
    <mitre>
      <id>T1190</id>
    </mitre>
    <group>sqli,web_attack,</group>
  </rule>
</group>

Mitigation & Patch Guidance

Patch confidence: 0/4 (UNPATCHED — no patch evidence from any tracked source).

1. Immediate mitigations

• Restrict network access to the /WebAgenda/ path via WAF, reverse-proxy ACLs, or firewall rules so that only trusted administrative networks can reach SMBAjaxConfigProcess.do.
• Deploy virtual patching rules (WAF/gateway) that block requests carrying SQL metacharacters in the id parameter.
• Enable verbose authentication in front of the WebAgenda module if the platform supports it; do not rely on default configurations.
• Review database account privileges used by Agentflow; ensure the account cannot execute OS-level procedures (e.g., xp_cmdshell where applicable) and cannot write outside application schemas.
• Monitor for anomalous SQL activity and unexpected data reads/writes originating from the Agentflow web tier.

2. Patch / upgrade steps

• No official patch reference is available in tracked sources. Contact Flowring support directly to request a fixed build for Agentflow 4.0 dated on or after 2025/08/08, and to confirm whether the flaw is remediated in that release.
• Until vendor confirmation is obtained, treat the 2025/08/08 build date in the CNA title as the presumed fix boundary and validate the specific id parameter handling yourself (code review or authorized penetration test).
• Track the NVD entry for CVE-2026-96429 for updates to patch references.

3. Long-term hardening

• Enforce parameterized queries / prepared statements throughout the application for all user-supplied input.
• Apply input validation and output encoding at every trust boundary; do not rely on client-side checks alone.
• Adopt a defense-in-depth posture: WAF, least-privilege DB accounts, network segmentation for internal workflow platforms, and centralized logging of web-tier SQL errors.
• Integrate dependency and CVE monitoring for Flowring Agentflow so future advisories are actioned quickly.

References

• NVD — CVE-2026-96429: https://nvd.nist.gov/vuln/detail/CVE-2026-96429
• MITRE CWE-89 — Improper Neutralization of Special Elements used in an SQL Command: https://cwe.mitre.org/data/definitions/89.html
• MITRE ATT&CK T1190 — Exploit Public-Facing Application: https://attack.mitre.org/techniques/T1190/
• FIRST EPSS (not yet scored for this CVE): https://www.first.org/epss/
• CISA Known Exploited Vulnerabilities Catalog (CVE-2026-96429 not listed at time of writing): https://www.cisa.gov/known-exploited-vulnerabilities-catalog

No known threat actor campaigns or exploit-in-the-wild evidence were identified in available data for CVE-2026-96429. No public PoC, Nuclei template, or exploit code has been published.