This CVE reveals a structural flaw in n8n's credential handling that extends well beyond the GraphQL node. The vulnerability isn't a missing sanitization call — it's a collision between two design assumptions. The credential subsystem assumes that controlling access during node execution controls the exposure surface. The execution persistence subsystem treats error payloads as debug metadata, not data requiring sanitization. When the HTTP client throws with headers intact and the node re-throws without filtering, the persistence layer stores the decrypted credential as a permanent artifact.
The CVSS base score obscures a practical reality: most n8n deployments grant execution history access more broadly than credential management access, since debugging workflows requires seeing execution state. If your deployment allows users to read execution history without credential admin rights, the effective exploit threshold is lower than the CVSS suggests. This isn't theoretical — execution artifacts surface in workflow status dashboards, debug interfaces, and audit trails that may be accessible to parties who never had credential management access.
The immediate action is patching the GraphQL node, but the deeper question is whether n8n's architecture can enforce credential sanitization across all error paths in all nodes. The fix pattern — wrapping errors in n8n's standard error type — is appropriate for this case, but it's a developer discipline solution to a design problem. Every future node author must remember this constraint in every error path forever. The structural condition likely exists elsewhere: decrypted credentials existing in memory alongside error objects that may escape sanitization. Treat execution artifacts as potentially containing secrets by default, and audit which users have execution read access without credential management rights.