Public Record

Federal Submissions

NIST CAISI RFI · Docket NIST-2025-0035

Security Considerations for AI Agents

Six policy recommendations for sealed governance artifacts, continuous measurement, tiered verification, and offline verifiability.

March 4, 202612 pagesPDF

NCCoE Concept Paper Response

AI Agent Identity & Authorization

Cryptographic identity binding, continuous runtime authorization, and offline-verifiable evidence bundles.

March 4, 202612 pagesPDF

Architectural Brief

Recent Analysis

Project Glasswing × AGA

An Architectural Complement

Direct mapping of the four AGA mechanisms to the three structural limits of behavioral monitoring acknowledged in Anthropic's Mythos Preview system card. MCP-native, independent audit primitive. Includes downloadable evidence bundle for offline verification.

May 1, 2026~7 min readWeb brief

Independent Research