Enkrypt AI MCP Scanner is a red teaming platform designed to identify vulnerabilities in AI systems including agents, RAG systems, tools, and Model Context Protocol (MCP) servers. The platform tests across multiple modalities including text, audio, and vision inputs. The tool performs automated security testing for prompt injection attacks (direct and indirect), jailbreaks, tool misuse, privilege escalation, data exfiltration, secrets leakage, and policy violations. It covers agent-specific failure modes such as goal hijacking, rogue agent behavior, cascading failures, and insecure inter-agent communication. Testing capabilities include RAG retrieval poisoning, memory poisoning, multimodal prompt smuggling, and agentic supply chain vulnerabilities. The platform generates findings with severity ratings, reproduction steps, and remediation guidance. Outputs include executive red team reports, findings registers with suggested fixes, regression test suites for CI/CD integration, and coverage maps showing tested surfaces. The platform supports compliance mapping to frameworks including NIST, OWASP, and EU AI Act. Users can configure testing through an SDK and API, with support for scheduled and on-demand testing in staging and production environments. The platform tests identity and privilege abuse scenarios, role bypass attempts, and tenant crossover attempts. Results can be exported to ticketing systems, SIEM platforms, and GRC workflows.