The vulnerability results from a missing authentication check for the --api-key flag when running the PraisonAI server in serve mode. Because the internal routes _create_agents_app() and _create_unified_app() do not enforce credential validation, any caller can send POST requests to /agents or /api/v1/agents/{id}/invoke. This allows an attacker to create or invoke agents without authorization, potentially leading to arbitrary code or command execution within the system and compromise of confidential data. The weakness falls under Missing Authorization (CWE-862).

Affected systems are installations of the PraisonAI multi‑agent platform by MervinPraison. All releases older than 4.6.58 are vulnerable because the fix was introduced in that release. If your version is not 4.6.58 or newer, the server exposes the vulnerable endpoints.

The CVSS score of 8.8 classifies this issue as high severity, and the lack of an EPSS value means we do not have a quantified modern exploitation probability, but the fact that the flaw permits unauthenticated use seriously elevates risk. The vulnerability is not listed in the CISA KEV catalog. Attackers can exploit this flaw by sending HTTP POST requests to the affected endpoints from any host that can reach the service, so network exposure increases risk. Because the flaw directly bypasses authentication, it likely requires neither local host access nor privileged credentials to exploit.