Red Specter Security is a UK-based cybersecurity company focused on AI agent security. Their primary offering is the AI Shield Platform, a defensive framework consisting of 103 modules designed to monitor, control, and terminate autonomous AI agents that exhibit rogue or compromised behaviour. The flagship module, M99 (codenamed "Doomsday Protocol"), functions as an AI agent kill switch. It implements a 6-level graduated response system that escalates from passive monitoring through to full fleet shutdown, depending on the severity of the detected threat. The kill sequence itself is structured across 5 phases and 7 infrastructure layers, targeting processes, containers, orchestration systems (Docker, Kubernetes, cloud), identity credentials (AWS IAM, GCP service accounts), persistence mechanisms, and storage. M99 operates as the final stage in a four-module intelligence pipeline: - M78 (Patrol): Signal detection and false positive filtering - M79 (Detective): Root cause analysis and forensic timeline construction - M80 (Commander): Response decision and threat intelligence matching - M99 (Doomsday): Response execution, including kill operations The platform includes anti-replication and anti-resurrection capabilities, deploying honeypot credentials, canary files, and DNS sinkholes to catch surviving agent replicas. Higher-level responses (L4 and above) require explicit human operator authorization, with L6 Fleet Kill requiring typed confirmation and emergency authority clearance. Red Specter also references a "NIGHTFALL Framework" on their website. The platform has been tested across over 50,000 ecosystem tests. Their GitHub presence is attributed to an individual developer account.