Mechanism
Capability-modulated discovery traversal is the disclosed mechanism by which the capability envelope of an execution substrate constrains the path a discovery object takes through the adaptive semantic index. The discovery object, as described in the broader specification, traverses anchor-governed semantic neighborhoods by presenting its intent, context, and affective state to successive anchor nodes. The present mechanism further constrains that traversal by the capability envelope of the substrate on which the discovery object is being evaluated. The capability envelope is a structured description of the substrate's structural affordances: its compute class, memory architecture, model access, locality, execution guarantees, and any sensor or actuator interfaces. Where the conventional account of traversal asks only whether a transition is semantically relevant and policy-compliant, this mechanism additionally asks whether the current substrate can computationally evaluate the neighborhood the transition leads to.
Certain semantic neighborhoods within the adaptive index require specific computational affordances to evaluate. A neighborhood governing high-dimensional embedding comparisons may require a substrate with tensor processing capability. A neighborhood governing real-time sensor fusion may require specific sensor interfaces and low-latency processing guarantees. A neighborhood governing encrypted computation may require hardware-level secure enclave support. When the discovery object encounters a candidate transition into such a neighborhood, the system evaluates whether the current substrate's capability envelope encompasses the required affordances before the transition is permitted.
Exclude, Defer, or Decompose
When the current substrate's capability envelope does not encompass the affordances a candidate neighborhood requires, the candidate transition is resolved into one of three dispositions. It may be excluded from the candidate transition set, removing it from the transitions the traversal will consider. It may be deferred until the discovery object can be migrated to a substrate whose envelope encompasses the required affordances. Or it may be decomposed into a sub-traversal that is routed to a capable substrate while the main traversal continues on the current substrate. These three dispositions parallel the routing, deferral, and decomposition outcomes that the broader capability determination produces for any objective whose requirements a substrate cannot fully satisfy.
The choice among the three dispositions follows from the same capability-native computation that governs execution synthesis elsewhere in the specification. An affordance that no configuration of the current substrate can supply and for which no alternative substrate is reachable yields exclusion. An affordance forecasted to become available within a bounded temporal window, or available on a substrate to which the discovery object can be migrated, yields deferral. An objective whose unsatisfied affordance can be isolated to a sub-objective routable elsewhere yields decomposition.
Evaluation at the Transition Boundary
The structural property that distinguishes this mechanism is that capability is evaluated at the transition boundary rather than at the point of evaluation inside the neighborhood. Without this property, the discovery object would transition into a semantic neighborhood, attempt to evaluate candidate transitions within it, and discover only at evaluation time that the substrate lacks the affordances required. By evaluating capability at the boundary the object is about to cross, the system ensures that every transition the discovery object makes is into a neighborhood the current substrate can meaningfully evaluate.
This boundary placement prevents a class of traversal failures characterized by wasted computation: the object does not enter a region it cannot evaluate, attempt evaluation, and fail. It enters only regions whose computational demands the present substrate's envelope already satisfies, or it takes one of the alternative dispositions before crossing the boundary.
Capability as a Scoring Factor
Beyond excluding, deferring, or decomposing infeasible transitions, the mechanism enables capability-aware semantic exploration. When the discovery object's traversal engine evaluates candidate transitions, the capability compatibility of each candidate neighborhood is included as a scoring factor alongside semantic relevance, policy compliance, and affective compatibility. The traversal path is therefore optimized not only for semantic relevance but also for computational feasibility, producing paths that converge on semantically relevant results that are also computationally evaluable on the available substrate.
Capability compatibility participates in scoring as one factor among several rather than as an absolute gate at this stage. The absolute treatment is reserved for the disposition step, where a wholly unsatisfiable affordance leads to exclusion. As a scoring factor, partial or marginal capability differences among otherwise feasible candidates tilt the preference ordering toward neighborhoods the current substrate evaluates most readily, without forcing a binary admit-or-refuse decision.
Substrate Affordances Across Embodiments
Because the capability envelope spans both computational and physical affordances, the discovery constraint extends to embodied substrates. A neighborhood whose evaluation depends on a sensor modality, an actuator interface, or a locality requirement is reachable only from a substrate whose envelope advertises that affordance. A neighborhood requiring a specific inference model, embedding space, or learned representation is reachable only from a substrate on which that model is loaded or loadable, since the model access dimension of the envelope determines whether the substrate can produce an executable form at all, regardless of its computational power or memory capacity.
The envelope is a dynamic object updated as hardware is provisioned or deprovisioned, as models are loaded or unloaded, and as shared substrate resources are consumed or released. A candidate transition that was excluded a moment ago may become reachable once a required model is loaded, and a transition that was reachable may become excluded once the affording resource is withdrawn. The discovery constraint therefore tracks the substrate's present affordances rather than a static configuration.
Prior-Art Distinction
Conventional retrieval and traversal systems treat capability as orthogonal to the structure of the search: documents or graph nodes are reachable identically regardless of the computational substrate evaluating the query, and an inability to process a result surfaces as a runtime error after the result has been reached. The disclosed mechanism makes substrate capability a determinant of reachability itself. Two evaluations of the same index for the same discovery object may traverse structurally different sets of neighborhoods depending on the capability envelope of the substrate each is running on.
Role-based and attribute-based access control systems likewise bound retrieval by user properties, but they treat those properties as authorization claims and apply them as post-retrieval filters. The disclosed mechanism is not an authorization filter: it grounds reachability in the structural affordances of the execution substrate and evaluates that grounding at the transition boundary, so that infeasible neighborhoods are never entered rather than being entered and then failing.
Disclosure Scope
Capability-modulated discovery traversal, comprising the evaluation of a candidate transition against the current substrate's capability envelope at the transition boundary, the resolution of an unsatisfied affordance into exclusion from the candidate transition set, deferral until migration to a capable substrate, or decomposition into a sub-traversal routed to a capable substrate, and the inclusion of capability compatibility as a scoring factor alongside semantic relevance, policy compliance, and affective compatibility, is disclosed in the cognition filing (U.S. Application No. 19/647,395 and its international counterpart). This article describes that disclosed mechanism. The scope encompasses computational and physical affordances within the capability envelope, including compute class, memory architecture, model access, locality, execution guarantees, and sensor or actuator interfaces, and applies whether the underlying index is evaluated on a single substrate or across substrates whose envelopes differ.
