Executive brief
next-ai-draw-io is a web application that integrates AI capabilities with diagramming tools. A security flaw allows remote attackers to forge user identities by manipulating network headers. This can lead to unauthorized use of expensive AI resources at the owner's expense, bypass of usage limits, and the ability to impersonate other users' activity logs.
Technical details
The vulnerability exists in the identity generation logic within `lib/user-id.ts`. The application derives the `userId` exclusively from the client-provided `X-Forwarded-For` HTTP header without verifying if the request passed through a trusted reverse proxy. Because this header is entirely client-controlled, an unauthenticated attacker can rotate header values to bypass DynamoDB-based quota enforcement (rate limiting) and impersonate other users by spoofing their IP addresses. This forged identity is also used for telemetry tracking in Langfuse and token usage accounting, leading to data corruption and potential financial exhaustion of the operator's AI API keys.
Affected products
- DayuanJiang next-ai-draw-io <= 0.4.13
Timeline
- 2026-03-16: other: Vulnerability discovered
- 2026-03-17: disclosed: Initial disclosure to vendor via GitHub issue
- 2026-07-21: advisory: CVE published