OAgents: A Behavioral Envelope Standard for Trustworthy AI Agent Operations — An Implementation Profile of the NIST AI Risk Management Framework (AI RMF 1.0)

The deployment of large language models as operational agents has outpaced the development of trust mechanisms for their behavior. This document proposes OAgents, a control and conformance profile for operational AI agent trustworthiness, presented as an Implementation Profile of the NIST AI Risk Management Framework (AI RMF 1.0, NIST AI 100-1). OAgents defines a behavioral envelope -- a structured set of pre-execution gates, post-execution verification, and operational discipline mechanisms -- comprising 26 components across 7 categories with three conformance levels and observable evidence criteria. The standard is model-agnostic: behavioral guarantees are properties of the envelope, not the underlying model. The profile is submitted as a community contribution in response to NIST's AI Agent Standards Initiative (CAISI, February 2026) and is consistent with NIST AI 100-1, NIST AI 600-1, and Executive Order 14110 on Safe, Secure, and Trustworthy Artificial Intelligence. A reference implementation demonstrating architectural feasibility of all 26 components is published at https://github.com/ologos-corp/OAgents-standard.