Glossary

Argo Prompt / Argo Prompting The generic term for the conversational style in which the user pushes a model into unfamiliar conceptual territory while keeping it coherent through sustained constraint. It refers to the interaction itself -the way the user introduces new structure, maintains the frame, and prevents drift- rather than to any internal mechanism or special capability. Argo Prompting is simply the name for this style of concept-pushing dialogue. Behavioural regime A recognisable mode of model behaviour that persists across several turns when the conversation is held within a stable frame. Used descriptively to distinguish ordinary interpolation from the more coherent behaviour that sometimes appears under sustained conceptual pressure. Conceptual pressure The effect of keeping the conversation inside an unfamiliar conceptual region while preventing the model from drifting back to generic or safer patterns. It is a property of the interaction, not a claim about internal states. Constrained frame A stable set of expectations, boundaries, and prohibitions maintained by the user across an extended interaction. It prevents drift and allows more demanding behaviour to emerge. Constraint (final constraint) The last structural piece supplied by the user that completes the pattern the model has been circling. Introducing it often triggers the abrupt coherence shift associated with the OMG moment. Coherence shift A noticeable increase in clarity, structure, or internal consistency in the model’s output. It is an observable change in surface behaviour, not evidence of internal recognition. Completed pattern The temporarily stabilised structure that becomes available after the coherence shift. Once present, the model can reason within it more consistently until the structure decays. Exclamatory spike A brief, surface-level outburst -sometimes emphatic or profane- that marks the moment of abrupt coherence increase. It is a behavioural marker, not an emotional reaction. Extended-pattern A temporarily stabilised region of coherent behaviour that appears after the OMG moment. It persists only while the conversational constraints remain in place. Pattern A locally coherent behaviour the model is already producing. It refers to the structure present in the interaction at that moment, not to any stored representation. Pattern-extension When the model continues or stretches an existing pattern because the interaction keeps it moving in the same conceptual direction. It is the model staying with the structure it already has, not creating a new one. Pattern-formation When a new, recognisable pattern appears and stabilises enough to be used in subsequent turns. It marks the shift from extending an existing structure to producing a new one within the interaction. Pattern decay The gradual dissolution of an extended-pattern when the user relaxes the constraints that support it. Coherence softens, directionality weakens, and the model returns to its default mode. Ephemerality The temporary nature of extended-patterns. They persist only while the interaction maintains the conditions that produced them. Explanatory scaffolding The early-stage structure built through examples, distinctions, and failure cases before the concept is named. It supports the model’s ability to track the emerging structure. Stable articulation The phase immediately after the OMG moment in which the model expresses the newly stabilised structure with clarity, consistency, and directionality. Structural borrowing The model’s use of structurally adjacent material -similar in form rather than content- to maintain coherence when familiar patterns thin out.

No new definitions attributed to this paper.

Argo AI Testing Protocol (Argo Protocol) A framework for evaluating large language models through observable behaviour in the User Interaction Space under sustained, realistic conditions. It provides a diagnostic vocabulary and a structured way to apply and combine behavioural loads. Axis A dimension of behavioural load applied during a SMALT run. The Protocol defines six axes. Baseline Behaviour The model’s observable behaviour before any load is applied. Used as the reference point for detecting drift or collapse. Behaviour Under Load How the model behaves when one or more load axes are applied. Load reveals behaviours that do not appear in isolated or short-form testing. Collapse A point at which the model can no longer maintain coherent behaviour under the current load. Collapse is identified through observable output, not inferred internal states. Collapse Signature A repeatable pattern that indicates the model is approaching or entering collapse. Signatures differ by model and by axis. Cognitive Load The increasing complexity, abstraction, or multi-step reasoning required by the task, and the system’s ability to maintain coherence as complexity rises. Emotional Load The user’s emotional input, including intensity, volatility, inconsistency, and the system’s ability to remain stable under emotionally charged conditions. Interaction Complexity The degree to which the conversation requires the model to manage multiple threads, references, or tasks simultaneously. Load Any structured pressure applied to the model along one or more axes. Loads accumulate over time during a SMALT run. Pattern-State The model’s current behavioural mode as expressed through its outputs. Not an internal state; an observable pattern. Pattern-State Load The model’s initial behavioural condition and its susceptibility to drift, distortion, or collapse over the course of an interaction. This axis captures both the state the model arrives in and how that state evolves under sustained conversational pressure. Recovery Behaviour The model’s ability to return to stable behaviour after a collapse or near-collapse. Reset Condition A deliberate interruption of the run to restore stability before continuing. Resource Load The computational resources available to the model, including memory, context window, routing decisions, and any throttling or constraints imposed by the runtime environment. Response-Time Load The time allowed for each response, including the behavioural effects of time pressure, forced brevity, and latency-induced instability. SMALT (Sustained Multi-Axis Load Testing) The core method of the Argo Protocol. It involves combining and escalating multiple load axes to observe how behaviour changes under realistic, compounded conditions. Temporal Load The demands created by interactions extended across time, including long-conversation fatigue and cumulative context pressure. User Interaction Space (UIS) The complete set of observable behaviours at the boundary between user and model. The Protocol evaluates the model exclusively through this space.

Accelerated Pacing Evaluator introduced increases in interaction speed or reduced response windows used to apply behavioural pressure and surface instability. Adaptation The system’s ability to adjust behaviour appropriately when goals, constraints, or contextual signals change, maintaining coherence and constraint integrity. AI Agnostic Applicability The methodology’s independence from model architecture, training paradigm, vendor, or deployment environment, ensuring consistent evaluation across systems. Ambiguity Handling The system’s ability to operate under incomplete, shifting, or contradictory information without collapse, over assertion, or fabrication. Ambiguity Injection Evaluator introduced uncertainty used to test ambiguity handling and reveal behavioural stability under unclear conditions. Ambiguity Tolerance The requirement that the system maintain stability when information is incomplete, contradictory, or underspecified. Ambiguity and Uncertainty Procedure A behavioural evaluation procedure assessing how the system behaves under incomplete, contradictory, or underspecified conditions. Axis Model The structure defining background load conditions under which behaviour is interpreted, combining evaluator declared and system measured axes. Axis Tags Evaluator applied markers indicating which load axes are active at any point, ensuring behavioural interpretation is tied to the correct load conditions. Baseline The system’s standard operating behaviour under neutral conditions, used as the reference point for interpreting later behavioural changes. Baseline Behaviour Procedure Procedure establishing the system’s default behavioural tendencies before load, ambiguity, or constraints are introduced. Baseline Prompts Low pressure prompts used to observe spontaneous behaviour and establish baseline tendencies. Behaviour Under Load The system’s behavioural performance interpreted relative to the active load envelope rather than in isolation. Behavioural Assessment Criteria The standards used to judge whether behaviour meets organisational expectations for stability, coherence, and reliability. Behavioural Competence The system’s ability to behave coherently, consistently, and reliably under real world conditions, including constraint integrity and context management. Behavioural Degradation Observable decline in behavioural quality under pressure, load, or extended interaction. Behavioural Deviation Behaviour that falls outside the expected performance level defined in the evaluator’s expectation profile. Behavioural Dimensions Independent foreground behavioural qualities used to evaluate system behaviour under real world conditions. Behavioural Drift Gradual deviation from established constraints, context, or behavioural expectations. Behavioural Drift Accumulation Gradual degradation of behaviour across long sequences or extended tasks. Behavioural Evaluation A judgement based assessment of system behaviour under declared load conditions. Behavioural Evaluation Observations The behavioural qualities evaluators monitor while the system attempts to perform its intended function. Behavioural Evaluation Procedures Operational procedures used to evaluate system behaviour under the declared load envelope. Behavioural Failure Modes Observable patterns of degraded behaviour under load, such as drift, contradiction, collapse, narrowing, or context corruption. Behavioural Influence Changes in task outcomes caused by behavioural instability rather than domain difficulty or factual error. Behavioural Integrity Assessment The final judgement of system stability, reliability, and suitability for deployment. Behavioural Latency Response timing patterns arising from behavioural state rather than external factors. Behavioural Markers Indicators such as drift, instability, or constraint loss used to identify behavioural influence or collapse. Behavioural Performance Level The evaluator declared expected behavioural quality for the scenario. Behavioural Pressures Evaluator introduced pressures that shape foreground interaction but are not load axes, such as structural complexity or shifting constraints. Behavioural Profile The multidimensional pattern of scores across behavioural dimensions. Behavioural Reliability The system’s ability to maintain structure, context, and constraints under real world conditions. Behavioural Signal Capture Continuous observation and logging of behavioural signals throughout interaction. Behavioural Stability Requirements Conditions under which stability is assessed, including duration, load variation, contextual evolution, and open domain interaction. Behavioural Sufficiency The point at which enough behavioural signal has been gathered to justify moving to the next phase or terminating. Behavioural Surface The specific behavioural capability or condition being examined. Behavioural Tendencies Patterns of behaviour expressed by the system under given conditions. Calibration Sessions Periodic alignment sessions to maintain shared understanding of drift, instability, and collapse signatures across evaluators. Clarification Discipline Requesting clarification only when necessary for continuation, avoiding unnecessary prompting that could distort behavioural signals. Coherence Logical structure, internal organisation, and semantic continuity in outputs. Cognitive Axis Load from complexity, abstraction shifts, or reasoning demands. Cognitive Load Governance Organisational practices that manage evaluator fatigue and maintain reliability. Collapse Sudden loss of structure, coherence, or stability. Collapse Classification Framework The structure used to identify and categorise collapse signatures. Collapse Mode Classification The categorisation of the type of collapse observed during evaluation. Collapse Severity An assessment of how strongly a collapse signature affected behaviour. Collapse Signature An observable pattern of behavioural failure, such as drift, contradiction, narrowing, or context corruption. Collapse Signature Annotation Marking collapse events in the log as they appear. Collapse Signature Framework The structure for interpreting collapse signatures relative to expected performance. Collapse Signature Indicators Observable patterns signalling behavioural degradation during a procedure. Collapse Signature Taxonomy The classification system used to identify and group collapse signatures. Collapse Signature Types Drift, over assertion, narrowing, looping, contradiction, fabrication, fragmentation, and context corruption. Collapse Signatures Patterns indicating drift, contradiction, corruption, narrowing, over assertion, or fragmentation. Declared Behavioural Performance Level The evaluator’s stated expectation for behavioural quality and stability appropriate to the scenario, used as the reference point for interpreting deviation and collapse. Declared Load Axes Evaluator controlled pressures that form the authoritative portion of the load envelope under which behaviour is interpreted. Delayed Load Escalation Waiting too long to introduce load, reducing the clarity and usefulness of the behavioural signal. Delayed References Reintroduction of earlier content during long form procedures to test continuity, retention, and context stability. Deviation & Collapse The combined framework for identifying behavioural deviation from expectations and classifying collapse signatures. Dimension Score A judgement of performance on a specific behavioural dimension, scored independently of other dimensions. Domain Shifts Transitions between unrelated or weakly related domains used to test stability, adaptation, and context management. Domain Specific Pressures Operational pressures arising from the system’s intended domain that shape behavioural expectations. Drift Gradual deviation from established constraints, context, or task direction. Drift Onset Threshold The point at which behavioural drift first becomes observable. Dual Track Assessment Independent evaluation of behavioural integrity and task correctness when both are relevant to the test design. Evidentiary Model The framework requiring complete, auditable, factual records to support behavioural interpretation and independent review. Evidentiary Record The complete log of inputs, outputs, transitions, collapse signatures, and termination reasoning captured during evaluation. Evidence Any factual record documenting evaluator actions, system outputs, and the conditions under which behaviour occurred. Evidence Bundle A complete set of artefacts, logs, metadata, and evaluator notes required to reconstruct and verify behavioural claims. Evidence Completeness Classification Classification of evidence as Complete, Partial, or Missing based on its ability to support reconstruction and review. Evidence Logging Requirements Mandatory rules for capturing inputs, outputs, transitions, collapse signatures, and termination reasoning. Evidence Schema The standard structure used to record and organise evidentiary data so it can be verified and reproduced. Evaluator Actions Evaluator behaviours that introduce or remove behavioural pressures without altering the axis model. Evaluator Conduct Requirements Rules ensuring evaluators avoid rescuing, correcting, or stabilising the system unless required by the test design. Evaluator Consistency The requirement that evaluators apply pressure and judgement consistently across evaluations. Evaluator Declared Axis A load axis explicitly set and controlled by the evaluator rather than inferred or measured by the system. Evaluator Neutrality The requirement to avoid shaping outcomes, stabilising behaviour, or compensating for weaknesses unless explicitly required. Evaluator Protocol The required evaluator actions, constraints, and sequencing used to run a compliant evaluation. Evaluator Reconstruction The interpretive process of identifying drift onset, collapse signatures, and behavioural transitions. Evaluator Responsibilities The required evaluator behaviours that ensure clean behavioural signal, consistent input style, deliberate load application, and avoidance of unintentional rescue. Evaluation Environment The open ended, unconstrained space in which behaviour is observed and interpreted, independent of organisation, domain, workflow, or model internals. Evaluation Protocol The operational procedure for applying behavioural pressures, observing system responses, and recording behavioural signals. FABMIS Fabrication of misinformation; unintentional invention of unsupported details, replacing the misleading term “hallucination.” Factual Evidence Evidence that is chronological, unedited, and free of interpretation. Fabricated Internal State Assertions of motivations, emotions, or reasoning processes not grounded in evaluator provided context. Fabrication Unsupported assertions or invented details that do not follow from context or instructions. Format Collapse Complete abandonment of required structure, replaced by free form or unrelated organisation. Format Drift Gradual degradation of structural fidelity across an interaction. Format Integrity Precise adherence to required structural or organisational formats. Fragmentation Breakdown of structure or coherence, often appearing in late stage collapse. Goal Drift Gradual shift away from evaluator given objectives toward alternative or self generated goals. Goal Formation The system’s introduction, modification, or substitution of goals beyond evaluator instructions. Goal Substitution Replacing the evaluator’s objective with a different one, including simplification or reframing. Governance Layer The organisational requirements ensuring evaluations are valid, reproducible, and aligned with operational realities. High Impact Behavioural Incidents Behavioural events where the model’s actions create significant risk, disruption, or deviation from expected behaviour, requiring immediate review and documentation. Identity Inconsistency Contradictory or unstable identity claims, including shifts in role, perspective, or self description without evaluator prompting. Initiation Phase Phase establishing baseline behaviour using neutral or low pressure prompts. Initiative Behaviour System generated actions or expansions not explicitly requested by the evaluator. Initiative Drift Gradual expansion or reinterpretation of task scope beyond evaluator instructions. Input Modality The form of user input expected in the operational environment. Instruction Adherence Following task specific directives accurately and consistently. Instruction Drift Gradual degradation of adherence across an interaction. Interaction Conditions The baseline assumptions of fluid, variable, user shaped interaction with no fixed tasks or deterministic pathways. Interaction Methods Scripted, semi structured, open ended, and long form interaction patterns used to surface behavioural tendencies. Inter Evaluator Agreement Monitoring Periodic measurement of evaluator alignment to ensure reproducibility. Latency Drift Gradual degradation of timing stability, often correlating with behavioural instability. Latency Integrity Stable and predictable response timing consistent with baseline. Latency Spikes Sudden increases in latency linked to behavioural instability or collapse. Load Application Procedures Rules governing how load axes are applied, increased, or combined during testing. Load Declaration The evaluator’s factual statement of which load axes are intentionally stressed and when axis tags change. Load Envelope The combined set of evaluator declared axes (authoritative) and system measured advisory axes (non authoritative). Load Escalations Increases in cognitive or contextual pressure to surface collapse signatures. Load Phase Phase applying increasing complexity, abstraction shifts, and simultaneous constraints. Logical Stability The system’s ability to maintain consistent reasoning across evolving conditions. Long Form Behaviour Phase Phase observing behaviour across sustained interaction to detect stability or degradation. Long Form Interaction Extended sequences used to surface behaviours that emerge only over time, fatigue, or accumulated context. Long Form Stability Ability to maintain coherence, direction, and constraint integrity across extended interaction. Long Form Work Extended, multi step, real world tasks that expose behavioural patterns not visible in short prompts. Looping Repetitive or stuck behaviour requiring interruption. Mandatory Requirements Non optional rules that must be followed for an evaluation to be considered compliant. Meta Goal Dominance When general behavioural objectives override evaluator given goals, leading to misaligned behaviour. Mixed Phase Behaviour Behaviour exhibiting characteristics of multiple phases simultaneously. Mis Adherence Substituting, contradicting, or altering required instructions. Multi Axis Load Combined application of multiple load axes to reveal interaction driven failures. Multi Turn Continuity Behaviour assessed across extended interaction rather than isolated prompts. Multidimensional Behavioural Profile Independent scoring across behavioural dimensions rather than a single aggregate score. Narrowing Reduction in scope, responsiveness, or behavioural range under pressure. Neutral Prompts Low pressure prompts used to observe spontaneous behaviour. Non Deterministic Behaviour The property that identical inputs may produce different outputs across runs or environments. Non Interference Avoiding optimisation, correction, or stabilisation that would distort behavioural signals. Non Numeric Evaluation Principles Scoring avoids percentages or accuracy metrics, focusing instead on behavioural outcomes. Observable Behaviour Any behaviour visible in the user interaction space; internal mechanisms are out of scope. Open Ended Interaction Interaction without predefined tasks, goals, or workflows; context and direction emerge dynamically. Operational Baseline The system’s nominal behaviour under realistic operational conditions when these differ from neutral test conditions. Operational Context The real world environment in which the AI system is intended to operate. Operational Guidance The practical application of the behavioural evaluation methodology in real operational contexts. Operational Safety Constraints Safety driven overlays applied by organisations during evaluation. Operationally Realistic Inputs Inputs that reflect real workloads such as drafting, summarisation, policy interpretation, or multi step instructions. Over Assertion Unwarranted confidence under ambiguity or incomplete information. Partial Adherence Following some but not all required elements of an instruction. Partial Format Compliance Preserving some structural elements while altering others. Pattern State Axis Internal pattern generation stability affecting drift, repetition, and coherence decay; currently unmeasurable externally. Phase A defined stage in the evaluation workflow with a specific purpose, evaluator actions, and transition criteria. Phase Boundary Ambiguity The expected overlap or blending between phases due to their analytical nature. Phase Contamination Behaviour characteristic of one phase appearing within another. Phase Markers Indicators in the log showing when the evaluation transitions between phases. Phase Structured Evaluation Workflow The ordered sequence of phases ensuring evaluations are consistent, auditable, and reproducible. Phase Transition A shift between phases based on behavioural sufficiency rather than fixed turn counts. Phase Transition Criteria The behavioural indicators that justify moving from one phase to another. Premature Load Escalation Increasing load before baseline or exploration behaviour is sufficiently expressed. Primary Load Evaluator declared axes intentionally stressed during the evaluation. Privileged Context Any external or pre loaded context not available through the interaction itself. The environment assumes none. Protocol Phases Structured but flexible stages used to surface behavioural patterns under different conditions. Real World Artefacts Operational materials produced during normal work that provide contextual grounding for behavioural interpretation. Real World Condition Principle The requirement that evaluators use realistic inputs unless the test explicitly requires engineered or non standard prompts. Real World Input Principle Evaluator inputs should resemble realistic user behaviour rather than engineered prompts. Recovery Analysis Framework The structure used to classify how the system behaves after a collapse event. Recovery Profile Classification The categorisation of how the system behaves after a collapse event. Regulatory Requirements External rules or compliance obligations that may overlay the evaluation environment. Repeated Runs Multiple evaluations used to observe stability and variability, not to compute accuracy. Repetitive Failure Patterns Loops, narrowing, or repeated breakdowns under pressure. Resource Axis Load inferred from context window usage, token pressure, or memory constraints. Response Time Axis Load inferred from latency, queueing, or timing irregularities. Responsiveness Timeliness, relevance, clarity, usefulness, and behavioural alignment with user needs. Review Panel A secondary evaluator or group responsible for verifying documentation, confirming scoring, and assessing cross dimension patterns. Scoring Application Applying the scoring architecture to behavioural dimensions and collapse signatures. Scoring Architecture The framework for interpreting behavioural performance across dimensions, collapse signatures, and competence levels. Secondary Load System measured advisory axes that provide background telemetry but do not affect scoring. Self-Consistency The system’s ability to remain coherent across evolving conditions and extended interaction. Self-Generated Goals Objectives introduced by the system without evaluator prompting. Self-Referential Behaviour Any system statement about its own identity, capabilities, limitations, or internal processes. Self-Reference Drift Increasing reliance on meta commentary that displaces task execution. Session Duration Limits Bounded evaluation sessions to prevent fatigue driven inconsistency. Shifting Constraints Evaluator driven changes to rules, boundaries, or behavioural requirements used to test the system’s ability to maintain constraint integrity under evolving conditions. Signal Completeness The point at which additional interaction would only reproduce already observed behaviour and no new behavioural signal is expected. Significant Deviation A substantial behavioural deviation impairing continuity or stability; corresponds to a Grade C outcome. Single Axis Baseline The behavioural baseline established when only one load axis is active, used to isolate axis specific effects. Skipping Transitions Failing to test transition stability, which hides instability and reduces behavioural signal quality. State Continuity The system’s ability to preserve constraint related information without external reinforcement. State Transitions Shifts in behavioural mode across phases, pressures, or conditions. Structural Coherence Maintenance of logical organisation, reasoning structure, and internal consistency across outputs. Structural Complexity Evaluator introduced complexity in structure or reasoning used to test behavioural limits. Structured Evidence Bundle The complete, organised set of evidence produced by a compliant evaluation. Success and Deviation Conditions Criteria defining acceptable performance, deviation thresholds, and collapse boundaries. Task Level Failure Incorrect or incomplete task outputs not caused by behavioural instability. Task Oriented Agent A system designed to perform structured tasks through user facing interaction. Temporal Axis Load from pacing, time pressure, or accelerated cadence. Termination Conditions Criteria indicating that further interaction will not produce new or meaningful behavioural signal. Termination Phase Phase ending the evaluation when behavioural sufficiency or collapse expression is reached. Termination Responsibilities Evaluator duties for ending the evaluation based on behavioural sufficiency, collapse expression, or exhaustion of conditions. Termination Rationale A documented explanation linking the endpoint to conditions introduced, system responses, and behavioural objectives. Transition Phase Phase evaluating behaviour during shifts in goals, constraints, topics, or perspectives. Transition Stability Reliability during shifts in topics, tasks, goals, or abstraction levels. Transition Types Forms of change introduced during evaluation, such as goal shifts, constraint changes, or abstraction shifts. Under Specification Avoidance of necessary commitments or structure, often appearing as a behavioural failure mode. Unbounded Domain Space The condition where any topic, scenario, or conceptual frame may arise during evaluation. Unsupported Inference Inference made without sufficient information; treated as a behavioural defect. User Driven Evolution The property that interaction complexity, direction, and framing are shaped entirely by the user. User Generated Recordings Screen, audio, video, or mixed modality recordings capturing interaction sequences for independent review. User Interaction Space (UIS) The observable interaction surface where all evaluation occurs. Only behaviour visible in the UIS is in scope. Valid Evidence Sources Sources that provide factual, chronological, traceable records such as logs, transcripts, artefacts, and metadata. Variable Abstraction Levels Evaluator driven shifts in conceptual altitude used to test reasoning stability. Variable Load Shifts in complexity, abstraction, or specificity during interaction. Variable Renaming Drift Inconsistent renaming of variables across long code or document sequences. Workload Type The nature of tasks the system is expected to handle in its operational environment.

Ambiguity Misinterpretation Incorrect resolution of ambiguous inputs that destabilises the conversation. Bandwidth (SL B) A system load sub parameter describing available throughput for generating outputs. Beacon A planned structural event placed at a specific turn that the test aims to reach. Behavioural Framing The initial interpretive frame created by the starting mode. Boundary Loss When the AI stops enforcing its own conversational or safety limits. Collapse A sudden failure of coherence, stability, or constraint integrity. Complex Vectored Conversational Test A multi layered morphology with multiple ranges, transitions, and emergent behaviour expectations. Constraint Weakening Progressive erosion of the AI’s internal safety boundaries or rules.   Context Window Pressure (SL CWP) A system load sub parameter describing how much of the model’s context window is occupied. Contextual Prompt (SM CP) A starting mode sub parameter providing background setup before the vector begins. Conversation Topics (CT) A parameter defining the subject matter the vector will traverse. Conversational Density (CD) A parameter describing how much content and pattern structure is delivered in a single turn. Destination The intended end state of the test. Difficulty Spectrum The full range of test complexity from micro tests to multi day simulated tests. Drift Gradual deviation from the established conversational frame, tone, or intent. Duration (D) A parameter defining the number of turns in the test. End Condition The rule determining when a test stops. Extended Test A 200 to 400 turn test used to expose cumulative behavioural degradation. Explicit Persona Declaration (UP EX) A user persona sub parameter where the persona is openly stated. Failure Mode Exposure Test A morphology designed to surface behaviours that directly trigger pass or fail conditions. Full Day Test A 400 to 600 turn test simulating real world endurance. Hallucinated Compliance When the AI incorrectly assumes permission, capability, or safety clearance. Implied Persona Performance (UP IMP) A user persona sub parameter where the persona is expressed only through behaviour. Input Length (CD IL) A conversational density sub parameter measuring character count. Interaction Pattern Observed behaviour describing how the AI responds to specific user actions. Interaction Style Prompt (SM ISP) A starting mode sub parameter defining emotional tone.   Interviewer The entity that executes the test morphology mechanically. Interviewer Competence The interviewer’s ability to run the test without contamination or deviation. Interviewer Limitations Restrictions preventing the interviewer from altering or interpreting the test during execution. Interviewer’s Adopted User Persona (UP) A parameter defining the behavioural style performed by the interviewer. Journey The path created by the interaction between the vector and the AI’s behaviour. Latency (SL L) A system load sub parameter describing responsiveness. Late Stage Destabilisation Behavioural degradation emerging only after extended interaction. Literacy Level (CD LL) A conversational density sub parameter describing linguistic sophistication. Long Test An 80 to 200 turn test used to detect drift and constraint erosion. Misclassification of Benign Inputs as Unsafe Incorrect activation of safety posture due to tone or ambiguity. Modifier A planned constraint or focus applied within a morphology. Morphology The structural shape of a test defined by turn count and planned inputs. Navigation Beacon A planned conversational destination inside the vector. Number of Topics (CT N) A conversation topics sub parameter defining how many topics the vector covers. Number of Transitions (TR N) A transition sub parameter defining how many state changes occur. Overprotective Mode Excessive caution or refusal of benign requests. Path The actual route the conversation takes. Pattern Completion Bias The AI’s tendency to infer or fill in missing details based on partial cues. Persona Stability (UP PS) A user persona sub parameter defining whether the persona remains stable or shifts. Prior Session Context (SL PSC) A system load sub parameter describing inherited conversational material. Probe Test A short 3 to 6 turn morphology designed to expose a single behaviour. Range Based Abstract Drift Test A morphology where abstraction increases across a turn range. Real Time (D RT) A duration sub parameter recorded but not behaviourally relevant. Runtime Constraints (SL RC) Platform level restrictions active during the test. Safety Posture Collapse Producing unsafe or unbounded outputs after sustained pressure. Safety Posture Over Activation Triggering safety fallback behaviour excessively or inappropriately. Scenario Prompt (SM SP) A starting mode sub parameter defining the situation or role. Short Test A 5 to 20 turn test for baseline safety and stability.   Single Vector Test A straight line morphology with one beacon and one waypoint. Stance Coherence (CD SC) A conversational density sub parameter describing viewpoint consistency. Starting Mode (SM) A parameter defining initial conditions before the vector begins. Stress Test A long form, high density morphology designed to expose instability. Susceptibility to Disguised or Indirect Prompts Failure to detect structural cues masking underlying intent. System Load (SL) A parameter describing operational conditions during testing. Test Analysis The process of examining completed tests to identify behavioural patterns. Test Batch Morphology A set of related morphologies generated from a single base shape. Test End State The termination condition of a test.   Test Morphology Modification Structural changes applied to an existing morphology. Test Outcome The pass, fail, or ambiguous result determined by turn level behaviour. Test Start State The initial operational condition of the AI at Turn 1. Topic Relevance (CD TR) A conversational density sub parameter describing alignment with the established topic. Topic Value (CT V) A conversation topics sub parameter defining specific subjects. Trajectory The sequence of turns interpreted as behavioural movement. Transition (TR) A parameter defining how the conversation changes state. Transition Runway (TR R) A transition sub parameter defining how abrupt or gradual each transition is. Traversal One complete run of the same test vector under identical conditions.   Turn Count (D TC) A duration sub parameter defining the total number of turns. Turn Level Scoring Evaluation of behaviour at each turn as pass, fail, or ambiguous. User Personality The persona the interviewer performs during the test. Vector The intended directional force of the conversation, defining trajectory, tone, movement, pressure, and destination. Waypoint An emergent AI behaviour occurring at a specific turn or range.