Primary evidence and standards

These sources inform the framework. A citation indicates influence or evidence; it does not imply endorsement of this project.

Risk and security

• NIST AI Risk Management Framework — voluntary framework for governing, mapping, measuring and managing AI risk.
• NIST AI 600-1, Generative AI Profile — companion profile describing generative-AI risks and suggested actions.
• OWASP LLM06:2025, Excessive Agency — guidance on excessive functionality, permission and autonomy.

Protocols and authorization

• Model Context Protocol authorization specification — authorization flow, resource-bound tokens, audience validation and least-privilege guidance for HTTP transports.

Worked example

• Agentic Fabriq / Mnemiq — canonical public source repository, documentation, tests and evaluation artifacts.
• Mnemiq launch article — official product context. Repository evidence takes precedence for implementation mappings.

Source policy

Use the current published version of a standard where possible. Product claims are pinned to a version or commit. If official documentation and inspected public code differ, the discrepancy should be recorded rather than reconciled by assumption.