CVE-2026-61426 describes a vulnerability in MervinPraison's PraisonAI product prior to version 1.7.3. The issue arises from an insecure default configuration that binds the service to all network interfaces, disables API key requirements, and enables wildcard CORS. This configuration allows unauthenticated remote attackers to access sensitive agent instructions and system prompts through the GET /api/agents endpoint and to invoke agents via the POST /api/chat endpoint without any authentication. The vulnerability has a CVSS 4.0 score of 8.8, indicating high severity due to network attack vector, no privileges required, no user interaction, and high impact on confidentiality.

An attacker can remotely access sensitive internal instructions and system prompts, potentially exposing confidential information. Additionally, attackers can invoke agent functionality without authentication, which may lead to unauthorized actions or data manipulation within the affected system. The vulnerability compromises confidentiality and partially impacts integrity, but does not affect availability or system control.

Patch status is not yet confirmed — check the vendor advisory for current remediation guidance. Until an official fix or update is released, users should avoid deploying PraisonAI with default configurations that bind to all interfaces, disable API key requirements, or enable wildcard CORS. Restrict network exposure of the service and implement authentication controls to prevent unauthorized access.