Mechanism
The composite admissibility evaluator is the governance gate through which every proposed mutation to the semantic agent's state must pass before commitment. As disclosed in Chapter 5 of the cognition specification, the confidence governor, the integrity engine, and the capability envelope operate together as a composite admissibility evaluator that integrates signals from a plurality of cognitive domain fields, including the affective state field, the integrity field, the confidence field, the capability field, and the personality field, to produce a composite admissibility determination for each proposed mutation. The evaluator does not reduce admissibility to any single dimension. It requires concurrent satisfaction of confidence sufficiency, integrity compliance, and capability confirmation before a proposed mutation is admitted for execution.
The proposed mutation may originate from an external inference engine, from the agent's own forecasting engine as a promoted speculative branch, or from a delegation. Whatever its origin, it is registered as a candidate mutation, a proposed change to the agent's state that cannot be committed until it has passed the evaluator. The specification refers to this composite admissibility evaluation as the governance gate: the evaluation that integrates signals from the cognitive domain fields to produce an admissibility determination for each proposed mutation before commitment.
Concurrent Satisfaction Across Dimensions
The defining property of the evaluator is that admissibility is not awarded on the strength of any one dimension. The specification states that the composite admissibility evaluator does not reduce admissibility to any single dimension and requires concurrent satisfaction of confidence sufficiency, integrity compliance, and capability confirmation. An agent that is highly capable but whose confidence has fallen below the execution authorization threshold is not admitted. An agent that is confident and capable but whose integrity field has degraded below its policy-defined threshold is not admitted. Each contributing subsystem evaluates its own dimension, and the composite determination is granted only when the required dimensions are concurrently satisfied.
This concurrency is enforced by the constituent subsystems acting as independent evaluators. The integrity engine computes the projected impact of a proposed mutation on each of its three integrity domains, and if the projected impact would cause the composite integrity score to fall below a policy-defined threshold, the mutation is flagged for enhanced scrutiny and the integrity impact assessment is supplied to the governance gate as an input. The capability envelope system confirms that the agent's current substrate supports the structural requirements of the selected branch. The confidence governor evaluates whether the agent has sufficient self-assessed readiness to proceed. The composite determination consumes all of these inputs together.
The Confidence Governor as a Hard Gate
Within the composite evaluator, the confidence governor occupies a distinguished role: it is a hard gate, not an advisory module, a monitoring dashboard, or a soft constraint that the agent may override through urgency or intent priority. The specification states that when the confidence governor determines that execution authorization should be withdrawn, execution ceases, and that the agent cannot override the withdrawal through self-assessment, affective escalation, or policy reinterpretation. Execution authority is granted by the confidence governor and revoked by the confidence governor, and no alternative pathway to execution exists that bypasses this gate.
The confidence governor evaluates the agent's confidence field, a first-class computed state variable that encodes the agent's assessed sufficiency to continue executing its current task given the agent's present internal state and the current state of the task and environment. The confidence computation integrates capability sufficiency from the capability domain, integrity state from the normative alignment domain, affective modulation from the affective domain, and environmental conditions from the context. If confidence falls below the execution authorization threshold, the eligible branches are not admitted and the agent transitions to a deliberative mode rather than committing the mutation. Because confidence is itself computed from capability, integrity, and affective inputs, the confidence governor functions as one face of the same composite evaluation rather than as a separate, competing test.
Position in the Mutation Lifecycle
The composite admissibility evaluator is not a single monolithic step but a set of governance evaluations distributed across the agent's mutation lifecycle, the sequence of cognitive operations from receiving a stimulus to deciding whether and how to respond. In the lifecycle disclosed in the synthesis chapter, the integrity engine projects the integrity impact of the proposed mutation, the forecasting engine generates and prunes speculative branches, the confidence governor evaluates self-assessed readiness for the eligible branches, and the capability envelope system confirms substrate support for the selected branch. A candidate output is committed only after it has passed integrity impact verification, confidence authorization, and capability confirmation.
The evaluator therefore sits between the proposal of a candidate mutation and its commitment as a governed state transition. Upstream, the forecasting engine and inference engine produce candidate mutations together with their supporting state. Downstream, commitment applies the admitted mutation and extends the agent's auditable history. Every state-affecting operation passes through this composite gate, and the gate consults the agent's full cognitive state rather than a single learned score.
Lineage-Recorded Determinations
Each admissibility determination is recorded in the agent's lineage. At commitment, the lineage records the candidate output, the admissibility determinations rendered, the integrity impact projections, the confidence state at commitment time, and the affective state at commitment time. The commitment is not merely the emission of an output; it is a governed state transition that extends the agent's auditable history, enabling after-the-fact verification that every governance gate was applied and every cognitive domain contributed its evaluation.
The specification establishes that the composite admissibility evaluator governs more than ordinary action commitments. The decision to deprioritize a lineage entry under governed forgetting is itself evaluated by the composite admissibility evaluator before the deprioritization is applied. In multi-agent resource negotiation, each offer, counteroffer, acceptance, and commitment is a mutation evaluated through the composite admissibility evaluator and recorded in the negotiating agent's lineage. The periodic re-evaluation of an agent's narrative identity is likewise evaluated by the composite admissibility evaluator and recorded in lineage with full provenance. The evaluator is the common gate through which these heterogeneous governed mutations pass.
Distinction From the Decision Evaluation Module
The specification draws an explicit architectural distinction between the composite admissibility evaluator and the structured decision evaluation module of the forecasting engine. The composite admissibility evaluator evaluates a single proposed mutation against governance criteria, producing a permit-or-deny determination for that mutation. The decision evaluation module, by contrast, activates when the planning graph contains two or more mutually exclusive candidate branches at a common decision point, and it evaluates a plurality of competing candidates against each other to produce a governed selection among alternatives.
These are complementary, not redundant. The decision evaluation module selects which branch the agent should pursue; the composite admissibility evaluator then determines whether the selected branch may be committed. The decision evaluation module is also distinct from the branch classification mechanism, which classifies branches individually as eligible, introspective, delegable, or pruned. The composite admissibility evaluator operates on a candidate mutation that has already survived branch classification and, where applicable, decision evaluation.
Distinction From Prior Art
Conventional autonomous agent systems, including runtime environments that provide pause and resume capabilities, suspend execution reactively in response to external failures or resource interruptions. The composite admissibility evaluator instead evaluates admissibility proactively and continuously from the agent's own computed cognitive state, enabling the agent to withhold a mutation before damage occurs rather than recovering after damage has occurred. Conventional systems also tend to reduce admissibility to a single check, such as a resource-availability test at dispatch time or a learned classifier output compared against a threshold.
The specification distinguishes the present approach by requiring concurrent satisfaction across independent cognitive dimensions, by maintaining capability and governance evaluation in architecturally separate subsystems with no bidirectional dependency so that capability evaluation is not contaminated by governance state and governance evaluation is not contaminated by capability state, and by recording each determination in lineage so the determination is a first-class auditable artifact. The hard-gate character of the confidence governor, which cannot be overridden by self-assessment, affective escalation, or policy reinterpretation, completes the distinction: admissibility is a structural property of the architecture, not an advisory signal the agent may disregard.
Disclosure Scope
The composite admissibility evaluator, comprising the integration of signals from a plurality of cognitive domain fields including the affective state field, the integrity field, the confidence field, the capability field, and the personality field to produce a composite admissibility determination for each proposed mutation, the requirement of concurrent satisfaction of confidence sufficiency, integrity compliance, and capability confirmation, the hard-gate character of the confidence governor that no alternative pathway to execution may bypass, and the recording of each determination in the agent's lineage, is disclosed in the cognition filing (U.S. Application No. 19/647,395 and its international counterpart) in Chapter 5 and synthesized in Chapter 14. This article describes that disclosed mechanism. The scope extends to embodiments in which additional cognitive domain fields contribute signals to the composite determination, and to the application of the evaluator to governed mutations beyond ordinary action commitments, including governed forgetting, resource negotiation, and narrative identity re-evaluation, provided the composite determination integrates plural cognitive dimensions and is recorded in lineage.
