CVE-2026-60087 describes an incorrect authorization vulnerability in MervinPraison's PraisonAI product prior to version 1.6.78. The product caches tool approval decisions solely by tool name, which enables attackers to bypass intended authorization checks by reusing approvals from benign operations to perform unauthorized, potentially harmful file write operations with arbitrary arguments in the same session. This flaw arises from insufficient validation of parameters after initial approval caching.

An attacker who obtains approval for a benign operation can exploit this vulnerability to execute dangerous file write operations with unreviewed parameters, potentially leading to unauthorized modification of files. This could compromise system integrity or lead to further exploitation depending on the environment where PraisonAI is deployed.

Patch status is not yet confirmed — check the vendor advisory for current remediation guidance. Until an official fix is available, restrict access to PraisonAI to trusted users only and monitor for suspicious activity involving tool approvals. Avoid approving operations unless fully verified. No vendor advisory or patch information is currently provided.