Answers are grounded in this paper and cite their sources.
Long-running agent workflows can lose task scope, source custody, authority, current environment state, unresolved conflicts, or knowledge of what another worker is doing. Operational Epistemic Situational Awareness (ESA) is proposed as the capability to maintain and query an evidence-qualified model of those conditions and to govern continuation when that model becomes stale, conflicted, missing, or outside authority. The architecture uses a governed meaning substrate as an orchestration sidecar, but a sidecar is not awareness by itself. Agent check-ins can be false or stale; receipts can record claims without proving effects; exclusive task claims can deadlock; handoffs can be unaccepted; and unmediated tools can bypass the substrate. The system must distinguish asserted, observed, inferred, stale, conflicted, and unknown state; combine voluntary reports with independent observations; preserve liveness and recovery; and govern sensitive operational data. This paper defines an operational state model, check-in and claim leases, accepted handoff capsules, evidence postures, refusal and escalation outcomes, bypass and availability conditions, cold-boot orientation through the Governed Meaning Base, realization funnel, and comparative evaluation. It makes no claim of consciousness or privileged agent self-knowledge.
No takes yet. Share an insight, caveat, or question.
Adam Ableman (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: