CVE-2026-1470: n8n: Code Injection enables RCE

CRITICALPoC AVAILABLECISA: ATTEND

Published January 27, 2026

CISO Take

CVE-2026-1470 is a critical RCE in n8n's expression engine—patch immediately or take n8n instances offline. CVSS 9.9 with Scope:Changed means a single compromised n8n user account becomes full host compromise, including access to every credential, LLM API key, and downstream service configured in the platform. If n8n is part of your AI agent or automation infrastructure, treat this as an active incident until patched.

What is the risk?

Severity is maximally high. The CVSS vector (AV:N/AC:L/PR:L/UI:N/S:C) means any authenticated user—including trial accounts or shared service accounts—can exploit this remotely with no user interaction and no special skills. CWE-95 (Eval Injection) is a well-understood, easily weaponizable class. n8n is widely deployed as the orchestration backbone for AI agent pipelines, giving post-exploitation access to LLM APIs, vector databases, secrets, and enterprise integrations. No CISA KEV listing yet, but the exploit PoC is public via JFrog Research—assume active exploitation in the wild.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
n8n	npm	—	No patch
209.2KOpenSSF 7.116 dependentsPushed 4d ago44% patched~19d to patchFull package profile →

Do you use n8n? You're affected.

How severe is it?

CVSS 3.1

9.9 / 10

EPSS

20.4%

chance of exploitation in 30 days

Higher than 97% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

Exploit Available

Exploitation: MEDIUM

Sophistication

Trivial

Exploitation Confidence

medium

○CISA SSVC: Public PoC

○Public PoC indexed (trickest/cve)

○EPSS exploit prediction: 20%

Composite signal derived from CISA KEV, VulnCheck KEV, CISA SSVC, EPSS, Metasploit, Exploit-DB, trickest/cve, Nuclei templates, and inthewild.io exploitation reports.

What is the attack surface?

AVNetwork

ACLow

PRLow

UINone

SChanged

CHigh

IHigh

AHigh

What should I do?

6 steps

PATCH
Apply the fix from commit aa4d1e5825829182afa0ad5b81f602638f55fa04 immediately—no patched version pinpointed in advisory, so pull latest n8n release and verify the commit is included.
ISOLATE
If patching is delayed, restrict n8n to internal network only; block external access to n8n UI and API. Disable self-registration.
ROTATE CREDENTIALS
Assume all secrets stored in n8n credentials vault are compromised—rotate LLM API keys, DB passwords, OAuth tokens, and webhook secrets.
AUDIT ACCOUNTS
Review n8n user accounts; remove any unnecessary users; enforce MFA where possible.
DETECT
Search logs for unusual process spawning from n8n process (child processes like bash, sh, cmd), outbound connections to unexpected hosts, or filesystem writes outside n8n data directory.
NETWORK CONTROLS
Egress-filter n8n host to only permitted destinations; alert on unexpected outbound connections.

What does CISA's SSVC say?

DecisionAttend

Exploitationpoc

AutomatableNo

Technical Impacttotal

Source: CISA Vulnrichment (SSVC v2.0). Decision based on the CISA Coordinator decision tree.

How is it classified?

Code ExecutionAuth BypassData ExtractionAgentFrameworkPluginAML.T0012- Valid Accounts AML.T0049- Exploit Public-Facing Application AML.T0050- Command and Scripting Interpreter AML.T0053- AI Agent Tool Invocation AML.T0072- Reverse Shell AML.T0081- Modify AI Agent Configuration AML.T0083- Credentials from AI Agent Configuration AML.T0085- Data from AI Services AML.T0105- Escape to Host

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Article 15-Accuracy, robustness and cybersecurityArticle 9-Risk management system

ISO 42001

6.1.2-AI risk assessment8.1-Operational planning and controlA.10.1-Information security vulnerability managementA.9.2-AI system security

NIST AI RMF

GOVERN 1.7-Processes for AI risk management are defined and executedGOVERN 6.2-Organizational teams commit to AI risk managementMANAGE 2.2-Mechanisms for detecting and monitoring AI risks are applied

OWASP LLM Top 10

LLM04:2025-Data and Model PoisoningLLM06:2025-Excessive AgencyLLM07-Insecure Plugin DesignLLM08-Excessive Agency

Frequently Asked Questions

What is CVE-2026-1470?

Is CVE-2026-1470 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2026-1470, increasing the risk of exploitation.

How to fix CVE-2026-1470?

1. PATCH: Apply the fix from commit aa4d1e5825829182afa0ad5b81f602638f55fa04 immediately—no patched version pinpointed in advisory, so pull latest n8n release and verify the commit is included. 2. ISOLATE: If patching is delayed, restrict n8n to internal network only; block external access to n8n UI and API. Disable self-registration. 3. ROTATE CREDENTIALS: Assume all secrets stored in n8n credentials vault are compromised—rotate LLM API keys, DB passwords, OAuth tokens, and webhook secrets. 4. AUDIT ACCOUNTS: Review n8n user accounts; remove any unnecessary users; enforce MFA where possible. 5. DETECT: Search logs for unusual process spawning from n8n process (child processes like bash, sh, cmd), outbound connections to unexpected hosts, or filesystem writes outside n8n data directory. 6. NETWORK CONTROLS: Egress-filter n8n host to only permitted destinations; alert on unexpected outbound connections.

What systems are affected by CVE-2026-1470?

This vulnerability affects the following AI/ML architecture patterns: agent frameworks, RAG pipelines, AI automation pipelines, LLM orchestration layers, data ingestion pipelines.

What is the CVSS score for CVE-2026-1470?

CVE-2026-1470 has a CVSS v3.1 base score of 9.9 (CRITICAL). The EPSS exploitation probability is 20.40%.

What is the AI security impact?

Affected AI Architectures

agent frameworksRAG pipelinesAI automation pipelinesLLM orchestration layersdata ingestion pipelines

MITRE ATLAS Techniques

AML.T0012Valid Accounts

AML.T0049Exploit Public-Facing Application

AML.T0050Command and Scripting Interpreter

AML.T0053AI Agent Tool Invocation

AML.T0072Reverse Shell

AML.T0081Modify AI Agent Configuration

AML.T0083Credentials from AI Agent Configuration

AML.T0085Data from AI Services

AML.T0105Escape to Host

Compliance Controls Affected

EU AI Act: Article 15, Article 9

ISO 42001: 6.1.2, 8.1, A.10.1, A.9.2

NIST AI RMF: GOVERN 1.7, GOVERN 6.2, MANAGE 2.2

OWASP LLM Top 10: LLM04:2025, LLM06:2025, LLM07, LLM08

What are the technical details?

Original Advisory

n8n contains a critical Remote Code Execution (RCE) vulnerability in its workflow Expression evaluation system. Expressions supplied by authenticated users during workflow configuration may be evaluated in an execution context that is not sufficiently isolated from the underlying runtime. An authenticated attacker could abuse this behavior to execute arbitrary code with the privileges of the n8n process. Successful exploitation may lead to full compromise of the affected instance, including unauthorized access to sensitive data, modification of workflows, and execution of system-level operations.

Exploitation Scenario

Attacker obtains valid n8n credentials via phishing, credential stuffing, or a shared/default service account. They create or modify a workflow and embed a malicious JavaScript expression (e.g., using require('child_process').execSync(...)) in a node's expression field. When the workflow executes, the expression evaluates outside the intended sandbox, spawning a reverse shell to attacker infrastructure. From there, the attacker extracts the n8n credentials vault—harvesting OpenAI API keys, database URIs, and OAuth tokens—then pivots to connected AI services, injects poisoned data into RAG vector databases, or exfiltrates training data and proprietary workflows. The entire attack chain from initial access to data exfiltration is achievable in under 30 minutes by a moderately skilled attacker.

Weaknesses (CWE)

CWE-95Improper Neutralization of Directives in Dynamically Evaluated CodePrimary

CWE-95 — Improper Neutralization of Directives in Dynamically Evaluated Code: The product receives input from an upstream component, but it does not neutralize or incorrectly neutralizes code syntax before using the input in a dynamic evaluation call (e.g. "eval").

[Architecture and Design, Implementation] If possible, refactor your code so that it does not need to use eval() at all.
[Implementation] Assume all input is malicious. Use an "accept known good" input validation strategy, i.e., use a list of acceptable inputs that strictly conform to specifications. Reject any input that does not strictly conform to specifications, or transform it into something that does. When performing input validation, consider all potentially relevant properties, including length, type of input, the full range of acceptable values, missing or extra inputs, syntax, consistency across related fields, and conformance to business rules. As an example of business rule logic, "boat" may be syntactically valid because it only contains alphanumeric characters, but it is not valid if the input is only expected to contain colors such as "red" or "blue." Do not rely exclusively on looking for malicious or malformed inputs. This is likely to miss at least one undesirable input, especially if the code's environment changes. This can give attackers enough room to bypass the intended validation. However, denylis

Source: MITRE CWE corpus.