Executive brief
Parse Dashboard, a management interface for Parse Server applications, contains a critical security flaw in its AI Agent feature. An unauthenticated attacker can remotely access a specific programming interface to read or modify any data within the connected databases. This could lead to a total compromise of customer data and application settings for organizations using the optional AI Agent functionality.
Technical details
The vulnerability exists in the AI Agent API endpoint (POST `/apps/:appId/agent`) due to a lack of authentication, missing CSRF validation, and insufficient authorization middleware. By chaining these flaws, an unauthenticated remote attacker can execute arbitrary database operations with 'master key' privileges. Additionally, a cache key collision between the master key and the read-only master key was identified. The issue affects the opt-in AI Agent feature in versions 7.3.0-alpha.42 through 9.0.0-alpha.7. A fix is available in version 9.0.0-alpha.8 which implements proper authentication and per-app authorization.
Affected products
- Parse Platform Parse Dashboard 7.3.0-alpha.42 through 9.0.0-alpha.7
Timeline
- 2026-02-19: patched: Version 9.0.0-alpha.8 released
- 2026-02-23: advisory: GitHub Security Advisory published
- 2026-02-25: disclosed: CVE published to NVD