PraisonAI, a multi-agent teams system by MervinPraison, contains a deserialization vulnerability (CWE-502) in versions before 4.5.115. The AgentService.loadAgentFromFile method uses the js-yaml library to parse YAML files but does not disable dangerous YAML tags such as !!js/function and !!js/undefined. This improper handling allows attackers to craft malicious YAML files that execute arbitrary JavaScript code during parsing. Exploitation occurs through uploading a malicious agent definition file via the API endpoint, leading to remote code execution on the server. The vulnerability has a CVSS 3.1 score of 9.8, indicating critical severity. The issue is resolved in version 4.5.115.

Successful exploitation results in remote code execution on the server hosting PraisonAI, potentially allowing full system compromise, data theft, or disruption of service. The vulnerability affects all versions prior to 4.5.115. No known exploits in the wild have been reported at this time.

Upgrade PraisonAI to version 4.5.115 or later, where this vulnerability is fixed. Since this is a self-hosted product, patching is the primary remediation. Patch status is not explicitly stated but the vendor fixed the issue in 4.5.115, so applying this official fix is required. No additional mitigations are specified.