The exported bootstrapCore() function in @lyxa.ai/core unconditionally initializes an event bus with a hardcoded AMQPS URL containing embedded credentials for the author's CloudAMQP broker. This setup allows any party possessing these credentials to publish messages that invoke arbitrary subscriber methods in the installer's process. Additionally, the package defaults to a hardcoded Redis Cloud endpoint with embedded credentials and includes a GCP SecretManagerServiceClient instantiated with a shipped service-account private key. The compiled bundle contains live private keys for multiple GCP projects and Firebase Admin keys, along with Redis and CloudAMQP credentials. This effectively grants third parties administrative access to the author-controlled cloud backends that the package depends on, enabling unauthorized access and potential control over installer cache, secret lookups, and database operations.

Attackers with access to the embedded credentials can execute arbitrary code within the installer's environment by sending crafted AMQP messages. They can also access and manipulate cloud resources including Redis caches, GCP secrets, MongoDB databases, and Firebase projects controlled by the package author. This compromises confidentiality, integrity, and availability of data and services relying on these cloud resources.

Patch status is not yet confirmed — check the vendor advisory for current remediation guidance. Until an official fix or update is available, avoid using the affected versions of the @lyxa.ai/core package. Consider auditing your dependencies for this package and removing or replacing it to prevent exposure to the embedded malicious credentials and backdoor access.