Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessObtain local access to toolkit host
-
ExploitSubmit crafted input containing SQL payload
-
ExecutionInjected query bypasses input handling
-
ImpactDisclose, tamper with data or cause denial of service
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A05:2025 - Injection
ASVS 5.0
V1 - Encoding and Sanitization Parameterized queries, prepared statements
MITRE ATT&CK
2 techniques across 2 tactics: Initial Access, Persistence The technique list with ATT&CK links and the kill-chain view is part of the Basic plan and up. |
| Exploitation Conditions | The CVSS vector requires local access (AV:L) with no authentication (PR:N), low attack complexity (AC:L), and no user interaction (UI:N), limiting exploitation to attackers who can reach the toolkit's input path on the local system rather than remotely over a network. Affected deployments are NeMo Agent Toolkit versions 1.4 through 1.6. The specific product feature, configuration, or interface that passes attacker-controlled input into the SQL query is not described in the available sources, so the precise trigger cannot be stated; the AV:L designation is the primary limiting factor and should be confirmed with NVIDIA, as it is atypical for a SQL injection flaw. |
| Impact | NVIDIA scores this 8.4 (High) with vector CVSS:3.1/AV:L/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H, and the key tension is the attack vector: the vendor assigned AV:L (local), yet SQL injection (CWE-89) is most commonly a network-reachable flaw - this discrepancy means the realistic risk hinges on how input reaches the vulnerable query path, and should be verified with NVIDIA. If AV:L is accurate, an attacker needs local access to the host running the toolkit, which materially lowers real-world exposure versus a remote SQLi despite the high base score; the offsetting factor is PR:N (no authentication) and AC:L (low complexity), so once local access exists, exploitation is straightforward with full C/I/A impact. No EPSS probability was provided, the CVE is not on CISA KEV, and no proof-of-concept or exploit reference appears in the input, so there is no public exploit identified at time of analysis and no evidence of active exploitation. The vuln.today decision label is this_week, indicating a prompt-but-not-emergency remediation priority consistent with a high-impact flaw that currently lacks exploitation signals. Net: a genuine high-severity bug for anyone running affected versions, but the local attack vector and absence of exploitation evidence place it below internet-facing critical issues - confirm the AV:L assumption, since a network-reachable SQLi would substantially raise priority. |
| Exploit ScenarioDerived | An attacker with local access to a host running NVIDIA NeMo Agent Toolkit (versions 1.4 through 1.6) supplies crafted input that is incorporated unsanitized into a backend SQL query, allowing the attacker to read or modify stored data or crash the service. The specific feature or interface that constructs the query is not described in the available sources. No public exploit identified at time of analysis. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Oct 28, 2026
SQL injection (CWE-89) in NVIDIA NeMo Agent Toolkit versions 1.4 through 1.6 lets an attacker who can reach the toolkit's input path on the local system inject arbitrary SQL, with potential for data tampering, denial of service, and information disclosure. NVIDIA rates the flaw 8.4 (High) and the assessed vector is CVSS:3.1/AV:L/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H - unauthenticated with no user interaction, but local, so exploitation requires local reachability of the vulnerable query path rather than network access; that AV:L designation is atypical for SQL injection and is the main factor shaping real-world risk. No public exploit code was identified at time of analysis and the issue does not appear in CISA KEV; per the current decision signal this is a this-week priority rather than an emergency.
RemediationAI
No vendor-released patch identified at time of analysis and no fixed version is stated in the supplied data - the EUVD range lists 1.4 to 1.6 as affected without naming a remediated build, so confirm the fixed release directly in NVIDIA's product-security advisory for 2026/5912 (https://github.com/NVIDIA/product-security/tree/main/2026/5912) and track the NVD and CVE.org entries for updates. Until a fix is published, the most defensible compensating control follows from the assessed AV:L vector: restrict local access to hosts, containers, and service accounts that run the toolkit, since exploitation requires reaching the vulnerable query path locally - the trade-off is reduced operational flexibility for multi-tenant or shared-development environments. Where the toolkit's data-backed features can be disabled or pointed at read-only, minimally privileged database credentials, do so; this caps the tampering, disclosure, and denial-of-service impact described in the advisory, at the cost of losing those features' functionality. Audit toolkit logs and database query logs for anomalous SQL from the toolkit's identity, and confirm with NVIDIA whether any network-facing configuration exists that could raise the attack vector above local, because the trigger path is not described in the available sources.
Affected VersionsENISA
Affected Packages
Source: OSV (Open Source Vulnerabilities)
Affects your stack
Affects 1 package in euvd
Checking this CVE against your own stacks and servers is part of the Basic plan and up.
Technical ContextAI · show details
NVIDIA NeMo Agent Toolkit is an NVIDIA framework for building and orchestrating LLM-based agent workflows, which commonly sit in front of data stores and retrieval backends. The assigned weakness is CWE-89 (Improper Neutralization of Special Elements used in an SQL Command), meaning attacker-influenced input is concatenated or otherwise passed into an SQL statement without adequate parameterization, letting the attacker alter query structure. The CPE cpe:2.3:a:nvidia:nemo_agent_toolkit:*:*:*:*:*:*:*:* and the EUVD range place affected builds at 1.4 to 1.6. Notably, the assessed vector sets AV:L with PR:N and UI:N - local access, no authentication, no interaction - which is unusual for a SQL injection and suggests the vulnerable query path is reachable from a local interface, embedded component, or locally invoked tool rather than from an exposed network endpoint. The specific feature, configuration, or interface that feeds attacker-controlled input into the query is not described in the available sources, so the precise trigger cannot be stated.
Know when Nemo Agent Toolkit gets its next vulnerability
- Watch products and get an email when something new is published
- Check this CVE against your own stack and servers
- Exploit and PoC references, threat intelligence from 22 sources
No credit card · 7-day full trial
Severity by source
AV:L/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
Retaining vendor AV:L as the described access model with PR:N since no authentication is indicated; SQL injection yielding disclosure, tampering, and DoS justifies C:H/I:H/A:H.
Primary rating from Vendor (nvidia).
CVSS VectorVendor: nvidia
DescriptionCVE.org
NVIDIA NeMo Agent Toolkit contains vulnerability where an attacker could cause a SQL injection issue. A successful exploit of this vulnerability might lead to data tampering, denial of service and information disclosure.
Executive ViewAI
A high-severity SQL injection vulnerability in NVIDIA NeMo Agent Toolkit versions 1.4 through 1.6 enables local users to inject arbitrary SQL, risking data tampering, denial of service, and information disclosure. Although no public exploit or active exploitation is known, the flaw affects a tool used for AI agent development, and its local attack vector means any user with local system access could abuse it. Leadership should prioritize remediation this week to protect sensitive data and maintain service availability.
No vendor-released patch identified at time of analysis. Interim mitigations: restrict local access to trusted users, monitor for anomalous SQL activity, apply input validation or query parameterization where feasible, and isolate affected systems from sensitive data.
Threat Intelligence Overview
How this decision was reached
The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.
What every action level means
SSVC AssessmentENISA
Priority Score
Attack Technique
SQL injection exploits the way applications construct database queries by mixing user input directly into SQL statements.
Information disclosure occurs when an application unintentionally exposes sensitive data that aids attackers in reconnaissance or directly compromises security.
Denial of Service attacks render applications or systems unavailable by overwhelming resources or triggering failure conditions.
EUVD-2026-110823
GHSA-j527-p23v-7mw6
NVD
CVE.org
EPSS
Exploit-DB