Executive brief
PraisonAI Platform, a framework for managing AI agents, contains a security flaw where it uses a publicly known "secret" key to secure user sessions by default. Because this key is hardcoded in the software's source code, an attacker can create their own digital keys to log in as any user, including administrators. This allows unauthorized individuals to view private data, modify settings, or delete entire workspaces and their contents.
Technical details
The praisonai-platform package uses a hardcoded JWT signing secret ('dev-secret-change-me') by default. While a 'production guard' exists in auth_service.py to prevent the use of this secret, the guard only triggers if the PLATFORM_ENV environment variable is explicitly set to something other than 'dev'. Since PLATFORM_ENV defaults to 'dev' and is not set in standard deployment configurations, the application remains vulnerable in production. An unauthenticated remote attacker can forge HS256 JWTs with an arbitrary 'sub' claim, gaining full access to the platform, including workspace owner privileges. This issue persists in version 0.1.4, which was previously intended to be the patched version. Version 0.1.6 is the recommended fix.
Affected products
- MervinPraison praisonai-platform <= 0.1.4
Timeline
- 2026-05-30: disclosed: Vulnerability discovered and reported to vendor.
- 2026-06-17: advisory: GitHub Advisory GHSA-f38v-77qj-h4jq published.
- 2026-06-17: patched: Fixed in version 0.1.6.