A critical authentication bypass in PraisonAI allows unauthenticated attackers to invoke agents and access sensitive system data.

This vulnerability, classified as CWE-287, occurs when the PRAISONAI_CALL_AUTH configuration is set to disabled, causing the verify_token function to fail. This allows an unauthenticated attacker to make unauthorized calls to the /api/v1/agents/{id}/invoke endpoint.

The exploitation of this flaw poses a significant risk to data confidentiality and integrity. By invoking agents without authentication, an attacker can gain unauthorized access to internal tools and private context, potentially leading to unauthorized data exfiltration or manipulation of agent logic. With a CVSS score of 8.2, this high-severity vulnerability warrants immediate attention to prevent unauthorized access to sensitive business workflows.

Immediate Action: Update the PraisonAI software to version 4.6.62 or later immediately. Ensure that the PRAISONAI_CALL_AUTH configuration is set to a secure, enabled state in your deployment environment.

Proactive Monitoring: Monitor API access logs for any suspicious requests directed at the /api/v1/agents/ endpoint, specifically looking for requests missing the required CALL_SERVER_TOKEN.

Compensating Controls: If immediate patching is not possible, implement a Web Application Firewall (WAF) rule to block or challenge all traffic to the /api/v1/agents/invoke endpoint until the patch can be deployed.

Public Exploit Available: Unknown.

Given the high CVSS score and the nature of the vulnerability, organizations must prioritize this update. The ability for an unauthenticated user to interact with internal AI agents represents a severe security gap that should be closed immediately by applying the vendor-provided patch and reviewing configuration settings across all production instances.