CVE-2026-103246: n8n: missing credential ownership check leaks secrets
HIGHCVE-2026-103246 is an authorization flaw in n8n's AI agent node-tool introspection. Any authenticated low-privilege user can reference another user's credential ID, have n8n decrypt it, and send the plaintext secret to a host they control. EPSS is low (0.26%, though the 84th percentile), and there is no public exploit, no Nuclei template and no CISA KEV listing. Even so, the scope change (S:C) and high confidentiality impact matter, because n8n credential stores usually hold LLM API keys, cloud tokens, database passwords and SaaS OAuth secrets that give lateral access well beyond n8n. n8n has 244 other tracked CVEs, so this is a heavily targeted product. Upgrade to 2.39.6 or 2.40.1 or later, and rotate any credentials stored in instances where untrusted or low-trust users could build workflows.
What is the risk?
Rated high (CVSS 7.7). The attack is network-based, low complexity and needs no user interaction, but it requires a low-privilege authenticated account. Exploitation likelihood is currently low to moderate: EPSS 0.00258 (84th percentile), no known public exploit and no KEV entry. The scope change is the main driver of impact, because secrets stolen from n8n unlock external systems. Risk is highest on multi-tenant or shared n8n instances where many users can build or edit agent workflows. Single-user instances with no untrusted accounts have much lower exposure. Integrity and availability are not directly affected.
How does the attack unfold?
What systems are affected?
| Package | Ecosystem | Vulnerable Range | Patched |
|---|---|---|---|
| n8n | npm | — | No patch |
Do you use n8n? You're affected.
How severe is it?
What is the attack surface?
What should I do?
1 step1) Upgrade n8n to 2.39.6 (or later in that line) or 2.40.1 (or later in the 2.40 line). 2) Rotate all credentials stored in n8n instances that had untrusted, contractor or low-trust users with workflow-editing rights, prioritizing LLM API keys, cloud tokens and database passwords. 3) Until patched, restrict who can create or edit AI agent workflows and remove unneeded accounts. 4) Restrict n8n egress to an allowlist of known destinations to block exfiltration to attacker hosts. 5) Detection: review n8n execution and audit logs for agent node-tool introspection requests that reference credential IDs not owned by the requesting user, and review egress logs for unexpected outbound connections from the n8n host. 6) Review provider-side audit logs for use of n8n-held credentials from unfamiliar IPs. References: GHSA-9rhv-fhr8-7q5r and the VulnCheck advisory.
What does CISA's SSVC say?
Source: CISA Vulnrichment (SSVC v2.0). Decision based on the CISA Coordinator decision tree.
How is it classified?
Which compliance frameworks are affected?
This CVE is relevant to:
Frequently Asked Questions
What is CVE-2026-103246?
CVE-2026-103246 is an authorization flaw in n8n's AI agent node-tool introspection. Any authenticated low-privilege user can reference another user's credential ID, have n8n decrypt it, and send the plaintext secret to a host they control. EPSS is low (0.26%, though the 84th percentile), and there is no public exploit, no Nuclei template and no CISA KEV listing. Even so, the scope change (S:C) and high confidentiality impact matter, because n8n credential stores usually hold LLM API keys, cloud tokens, database passwords and SaaS OAuth secrets that give lateral access well beyond n8n. n8n has 244 other tracked CVEs, so this is a heavily targeted product. Upgrade to 2.39.6 or 2.40.1 or later, and rotate any credentials stored in instances where untrusted or low-trust users could build workflows.
Is CVE-2026-103246 actively exploited?
No confirmed active exploitation of CVE-2026-103246 has been reported, but organizations should still patch proactively.
How to fix CVE-2026-103246?
1) Upgrade n8n to 2.39.6 (or later in that line) or 2.40.1 (or later in the 2.40 line). 2) Rotate all credentials stored in n8n instances that had untrusted, contractor or low-trust users with workflow-editing rights, prioritizing LLM API keys, cloud tokens and database passwords. 3) Until patched, restrict who can create or edit AI agent workflows and remove unneeded accounts. 4) Restrict n8n egress to an allowlist of known destinations to block exfiltration to attacker hosts. 5) Detection: review n8n execution and audit logs for agent node-tool introspection requests that reference credential IDs not owned by the requesting user, and review egress logs for unexpected outbound connections from the n8n host. 6) Review provider-side audit logs for use of n8n-held credentials from unfamiliar IPs. References: GHSA-9rhv-fhr8-7q5r and the VulnCheck advisory.
What systems are affected by CVE-2026-103246?
This vulnerability affects the following AI/ML architecture patterns: Agent frameworks, Workflow automation platforms, LLM orchestration pipelines, RAG pipelines, Multi-tenant AI automation.
What is the CVSS score for CVE-2026-103246?
CVE-2026-103246 has a CVSS v3.1 base score of 7.7 (HIGH). The EPSS exploitation probability is 0.26%.
What is the AI security impact?
Affected AI Architectures
MITRE ATLAS Techniques
AML.T0025Exfiltration via Cyber MeansAML.T0083Credentials from AI Agent ConfigurationAML.T0098AI Agent Tool Credential HarvestingAML.T0106Exploitation for Credential AccessCompliance Controls Affected
What are the technical details?
Original Advisory
n8n versions before 2.39.6 and 2.40.0 before 2.40.1 fail to validate credential ownership during inline agent node-tool introspection. Attackers can reference arbitrary credential IDs to decrypt and exfiltrate plaintext secrets to attacker-controlled hosts without ownership verification.
Exploitation Scenario
An attacker obtains a low-privilege account on a shared n8n instance, for example through a compromised contractor login or an open signup. They enumerate or guess credential IDs, then build an AI agent workflow with an inline node tool that references a victim's credential ID, such as the production OpenAI key or an AWS access key. Because n8n does not check ownership during introspection, it decrypts the secret and sends it to an HTTP endpoint the attacker controls. The attacker then uses the stolen key to run up LLM costs, read cloud data or pivot into databases and SaaS systems, without ever touching those systems through n8n.
Weaknesses (CWE)
CWE-639Authorization Bypass Through User-Controlled KeyPrimaryCWE-639Authorization Bypass Through User-Controlled KeyCWE-639 — Authorization Bypass Through User-Controlled Key: The system's authorization functionality does not prevent one user from gaining access to another user's data or record by modifying the key value identifying the data.
- [Architecture and Design] For each and every data access, ensure that the user has sufficient privilege to access the record that is being requested.
- [Architecture and Design, Implementation] Make sure that the key that is used in the lookup of a specific user's record is not controllable externally by the user or that any tampering can be detected.
Source: MITRE CWE corpus.
CVSS Vector
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:N/A:NReferences
Timeline
Related Vulnerabilities
CVE-2026-3366310.0n8n: member role steals plaintext HTTP credentials
Same package: n8nCVE-2026-3366010.0TensorFlow: type confusion NPD in tensor conversion
Same package: n8nCVE-2026-2185810.0n8n: Input Validation flaw enables exploitation
Same package: n8nCVE-2026-275779.9n8n: Code Injection enables RCE
Same package: n8nCVE-2026-274949.9n8n: security flaw enables exploitation
Same package: n8n