SAT-SA-with-PQC

Modern cybersecurity assessment generates massive amounts of complex evidence, which often can be very heterogeneous, and difficult for human examiners to find meaningful evidence of risk, anomalies, execution gaps, and coverage gaps, while also maintaining trustworthy evidence provenance. To address this problem, SAT-SA provides a 32-agent offline, evidence-driven supervisory analytics platform with Machine Learning, MLOps and Agentic AI which is coupled with an anomaly detection, drift analysis, threat analysis, correlation, risk scoring and prioritization. Specialized agents provide bounded, explainable recommendation of the evidence gathered in the security domain, with final decisions left to humans under supervisory control. The TRUST-SAT layer provides Post-Quantum Cryptographic assurance via ML-DSA, ML-KEM and SHA3-256 with a hash-chained evidence ledger which relates evidence and findings to risk assessments, recommendations and human decisions. It yields a human supervised, air-gapped security assessment system which produces evidence, reasoning context, and cryptographic verification for the trustworthiness of the assessment, in addition to the information it discovers that needs attention. Repository: https://github.com/PrathamKapoor/SAT-SA-with-PQC Live Demo: https://trustsat.mpst.me/ Demo Organization Roles and Credentials: Supervisor (can record decisions) Credential: b30f077adfb7b2c8.eT6lD1HY5_sUF6-DZC2bzsIhsC_9tL2nN1g8WN6cgnQ