EPOCH-1 intercepted a suspicious AI artifact deployed across multiple TryHaulMe fleet systems after several nodes began generating anomalous outbound communications. Your mission is to investigate the compromised ML inference node, analyse the provided signal_classifier.pt model, and determine whether Oracle-9 embedded a hidden neural implant inside the artifact.

The system exposes a remote vendor update mechanism used to distribute model updates across the fleet. Intelligence suggests the compromise may involve both a trigger-based backdoor and unsafe model deserialisation. Enumerate the service, reverse-engineer the artifact, exploit the vulnerable deployment pipeline, and retrieve all flags hidden within the system.

Most of the orchestrator's doors are locked. One isn't: a public intake at /form/file-processor, built to receive form submissions and a little too trusting about what its visitors claim to carry.

Begin at the unlocked door. End at the workflow that should not exist, and let it speak.