A vulnerability in FlowiseAI Flowise allows authenticated users to bypass input validation filters, potentially leading to unauthorized data manipulation or system compromise.

The application is susceptible to an incomplete list of disallowed inputs (CWE-184), which occurs when input validation mechanisms fail to block malicious payloads. This vulnerability requires an attacker to have authenticated access (Low Privileges) to the Flowise user interface.

The exploitation of this vulnerability can result in the compromise of sensitive data or unauthorized flow execution within the LLM environment. With a CVSS score of 8.7, this issue is rated as high severity, reflecting the significant potential for impact on the integrity and availability of the customized LLM flows managed by the platform.

Immediate Action: Update Flowise to version 3.1.3 or later to apply the necessary input validation patches.

Proactive Monitoring: Review application access logs for unusual patterns involving API requests or unexpected input strings that deviate from standard flow configurations.

Compensating Controls: Implement strict network access controls to limit access to the Flowise interface only to trusted, authenticated users.

Public Exploit Available: Unknown.

Given the high CVSS rating and the availability of a proof-of-concept, administrators should prioritize updating to version 3.1.3 immediately. Failure to patch allows authenticated users to potentially manipulate LLM flows, which could lead to severe operational disruption or data leakage.