Product
MaxKB AI assistant platform is vulnerable to unauthenticated remote command execution due to prompt injection flaws in the 'execute' tool and insufficient sandbox escaping protections.