Reasonable security threats and vulnerabilities are analysed for the intended operating context, covering the development environment, the supply chain, the data chain and any other threat vector; guards and mitigations give appropriate confidence that security is maintained through development and use for every identified vector; the AI design within its system minimises the adversarial attack surface so far as reasonably practicable; and all remaining known vulnerabilities are communicated to the risk owner for management.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.