witnora

NIST AI Agent Standards Initiative Mapping

Source: NIST AI Agent Standards Initiative and NIST’s February 2026 launch announcement.

NIST frames the initiative around trusted, interoperable, and secure agentic AI adoption. Witnora maps to this direction by producing portable evidence about agent behavior before release and at runtime.

Witnora Evidence Coverage

NIST direction	Witnora evidence
Secure agent behavior	MCPBench policy violations, tool-call traces, taint tracking, and risk classification.
Interoperable assurance	JSON schemas for result, evidence item, and evidence bundle artifacts.
Runtime accountability	Onegent Runtime approval, mock execution, verification, and audit packet records.
Testing and benchmark readiness	Deterministic MCPBench suites, Tripwire CI fault injection, failure scenario library.
Cross-framework posture	Framework-neutral CLI and schemas that can ingest multiple product artifacts.

Current Gaps

• No alignment with a final NIST control catalog because the initiative is still developing.
• No agent identity or signing layer yet.
• No formal integration with ISO/IEC or government procurement workflows.

Direction

Witnora should focus on evidence portability and repeatability so it can track future NIST agent standards without tying the project to one vendor framework.

This site is open source. Improve this page.