Executive brief
n8n is a workflow automation tool used to connect different software services. A security issue was found where sensitive information, such as API keys for AI services like OpenAI and Anthropic, is saved in plain text within the system's execution logs. Any logged-in user with access to these logs can see these secrets, potentially allowing them to hijack connected AI accounts or access private data.
Technical details
A sensitive information disclosure vulnerability (CWE-532) exists in n8n's LLM sub-nodes, including those for OpenAI, Anthropic, and Lemonade. While custom HTTP headers are masked in the user interface, the underlying application writes these values in plaintext into the execution data stored in the database. An authenticated attacker with low privileges who has access to workflow execution records can retrieve these secrets. The vulnerability is particularly impactful because execution data is often persisted long-term and can be exported. The issue is resolved in versions 1.123.64, 2.29.8, and 2.30.1.
Affected products
- n8n-io n8n < 1.123.64, >= 2.0.0-rc.0 < 2.29.8, >= 2.30.0 < 2.30.1
Timeline
- 2026-07-08: disclosed: Initial disclosure by Jubke
- 2026-07-22: advisory: GitHub Advisory published
References
- https://github.com/n8n-io/n8n/security/advisories/GHSA-89gh-3pgc-v5h2
- https://github.com/n8n-io/n8n/releases/tag/n8n@1.123.64
- https://github.com/n8n-io/n8n/releases/tag/n8n@2.29.8
- https://github.com/n8n-io/n8n/releases/tag/n8n@2.30.1
- https://www.vulncheck.com/advisories/n8n-before-credential-exposure-via-llm-node-execution-data