CVE-2026-34936 MervinPraison CVE debrief

Who should care

Users of PraisonAI versions prior to 4.5.90, particularly those responsible for managing and securing PraisonAI deployments, should be aware of this vulnerability. They should take necessary actions to protect their systems, review system logs for suspicious activity, and ensure that all necessary security measures are in place to prevent similar vulnerabilities in the future. This includes verifying the integrity of API requests and responses, as well as implementing robust logging and monitoring mechanisms to detect and respond to potential security incidents in a timely manner.

Technical summary

The passthrough() and apassthrough() functions in praisonai accept a caller-controlled api_base parameter that is concatenated with the endpoint and passed directly to httpx.Client.request() when the litellm primary path raises an AttributeError. No URL scheme validation, private IP filtering, or domain allowlist is applied, allowing requests to any host reachable from the server. This issue arises because the functions do not properly validate or restrict the api_base parameter, which can lead to unauthorized requests being made to any host that is reachable from the server. The vulnerability is patched in version 4.5.90, and users should update to the latest version to mitigate this vulnerability effectively.

Defensive priority

High, given the potential for unauthorized requests to any host reachable from the server, and the availability of a patch in version 4.5.90 or later. Users should prioritize updating PraisonAI deployments to mitigate this vulnerability effectively and monitor system logs for potential exploitation attempts post-patch application. Additionally, restricting access to the PraisonAI system and implementing compensating controls can help reduce the risk of exploitation until patches can be applied. It is also recommended to perform a thorough review of system configurations and ensure that all necessary security measures are in place to prevent similar vulnerabilities in the future. This includes verifying the integrity of API requests and responses, as well as implementing robust logging and monitoring mechanisms to detect and respond to potential security incidents in a timely manner. Furthermore, users should consider conducting regular security audits and penetration testing to identify and address any potential vulnerabilities before they can be exploited by attackers. By taking a proactive and multi-layered approach to security, organizations can help protect their PraisonAI deployments against this and other potential threats. The CVE record was published on 2026-04-03T23:17:05.853Z and was last modified on 2026-07-24T22:10:00.140Z. The NVD entry is currently Analyzed, providing additional context and details about the vulnerability and its potential impact on affected systems. Overall, the defensive priority for this vulnerability is High due to its potential impact and the availability of effective mitigations, including patching and compensating controls. Users are advised to take immediate action to protect their PraisonAI deployments and prevent potential exploitation of this vulnerability. This may involve coordinating with vendors, applying patches, and implementing additional security measures to ensure the integrity and confidentiality of sensitive data. By prioritizing the security of their PraisonAI deployments and taking proactive steps to mitigate this vulnerability, organizations can help reduce the risk of exploitation and protect their systems

Recommended defensive actions

• Update PraisonAI to version 4.5.90 or later
• Restrict access to the PraisonAI system
• Monitor PraisonAI logs for suspicious activity
• Review system configurations to ensure all necessary security measures are in place
• Verify the integrity of API requests and responses
• Implement robust logging and monitoring mechanisms
• Conduct regular security audits and penetration testing

Evidence notes

The CVE record was published on 2026-04-03T23:17:05.853Z and was last modified on 2026-07-24T22:10:00.140Z. The NVD entry is currently Analyzed. The vulnerability affects PraisonAI versions prior to 4.5.90. Users should verify their deployments and review official advisories for mitigation guidance.

Official resources