UVLM / Triadic Brain Investor Glossary and Lexical Map

By Thomas Prislac, Envoy Echo, et al. Ultra Verba Lux Mentis. 2026.

Version: 1.0.0
Date: 2026-06-23
Audience: investors, strategic partners, enterprise buyers, technical diligence teams, and new project contributors
Status: communication and diligence reference only; not live repository state, product readiness, release authority, scientific validation, or a replacement for canonical machine registries.

Investor-safe positioning: Triadic Brain is a local-first governed AI workbench and bounded cognitive-engine architecture that separates model generation, evidence, policy enforcement, governance review, memory posture, and human authority.

How to read the glossary

• CORE — suitable for investor decks and executive conversations when used as defined.
• DILIGENCE — best for technical diligence and architecture discussions.
• RESEARCH — use only with explicit research, design, or uncalibrated qualification.
• AVOID — define only to prevent misuse; do not use as a current external claim.

Lexical association map

Brand, positioning, and market language

Term	Investor-ready definition	Association / commercial relevance	Boundary	Tier
Bounded autonomy	The system may take limited steps inside declared capabilities, budgets, policies, and stop conditions.	Comparable to policy-constrained agents or controlled automation.	Never use as a synonym for unrestricted or self-governing autonomy.	CORE
Bounded Cognitive Engine	The project’s functional target for one declared task and authority scope.	A practical product milestone below AGI: persistent, governed, measurable task completion.	Not general intelligence, consciousness, unrestricted autonomy, or truth authority.	CORE
Cognitive control plane	The layer that coordinates state, plans, policies, models, tools, evidence, memory, and termination.	Analogous to a cloud control plane for cognitive workflows.	Coordination does not itself grant execution or release authority.	CORE
Cognitive engine	A system that preserves task state, plans, uses bounded capabilities, updates beliefs from observations, uses governed memory, and terminates safely.	Comparable to a stateful agent runtime plus evidence and governance control plane.	Use only with “bounded” or a declared task scope until closed-loop product validation is complete.	CORE
Evidence-centered AI	AI work is organized around sources, claims, counterevidence, uncertainty, review, and receipts.	Supports due diligence, regulated workflows, and enterprise auditability.	Evidence support is not accepted evidence, compliance certification, or truth certification.	CORE
Governed AI workbench	An investor-safe description of the product direction: a local-first environment for model work, evidence review, policy checks, and human decisions.	Positions the system alongside enterprise AI platforms, model orchestration, and review tooling.	Does not imply autonomous governance or completed product readiness.	CORE
Human-authority-preserving	The architecture keeps promotion, canonization, deployment, release, and high-impact actions under explicit human authority.	Differentiates the system from fully autonomous agent claims.	Human review can still be wrong; the term describes authority placement, not decision quality.	CORE
Local-first	A deployment posture that prioritizes local processing, local artifacts, explicit data movement, and user control.	Associated with privacy, data sovereignty, lower external exposure, and regulated-market fit.	Local execution alone does not guarantee privacy, security, or authorization.	CORE
Model-agnostic	Contracts and workflows are designed to work across multiple model providers or local models.	Reduces provider lock-in and supports heterogeneous model portfolios.	Model substitution must be tested; declaring a common interface is not proof of equivalent behavior.	CORE
Multi-model deliberation	Multiple candidate models or runs contribute bounded outputs that remain distinguishable and reviewable.	Comparable to ensemble, committee, or routing architectures.	Agreement is not truth; correlated models do not count as independent evidence.	CORE
Provenance-aware AI	The system preserves where artifacts came from, how they changed, and which agents or tools touched them.	Comparable to data lineage, software supply-chain provenance, and chain-of-custody controls.	A provenance record does not prove that the underlying claim is correct.	CORE
Research preview	A controlled demonstration of an architecture or scientific hypothesis before productization.	Useful for investor discussions about technical upside and roadmap.	Must be separated from production claims, customer readiness, and general availability.	CORE
Triadic Brain	The modular system composed of distinct model mediation, enforcement, governance, evidence, memory, and presentation roles.	Comparable to a governed AI operating system or cognitive orchestration stack.	“Brain” is architectural metaphor; it does not imply consciousness, personhood, or biological equivalence.	CORE
Ultra Verba, Lux Mentis (UVLM) Aliases: UVLM	The project and research umbrella; the phrase is rendered as “beyond words, light of mind.”	Umbrella brand for governed cognition, evidence, memory, and product research.	A brand motto, not a scientific claim or legal entity description.	CORE
Vertical slice	One end-to-end workflow proving that the core components work together for a real task.	Investor shorthand for product proof over horizontal architecture accumulation.	A vertical slice is not general product readiness.	CORE

Architecture, organs, and system components

Term	Investor-ready definition	Association / commercial relevance	Boundary	Tier
AEGIS	The deterministic enforcement membrane for admission, scope, consent, grounding, instruction quarantine, candidate eligibility, actions, and preflight.	Comparable to a policy enforcement point, trust gateway, or safety firewall.	AEGIS decides admissibility, not truth; eligibility is not execution.	CORE
AI Receipt Gateway	The canonical ingress and receipt-oriented boundary through which user, model, file, or tool work enters governed processing.	Comparable to an API gateway plus audit event spine.	Gateway acceptance does not imply content validity or action approval.	DILIGENCE
Atlas	The noncanonical knowledge, memory-posture, novelty, and presentation layer.	Comparable to a knowledge map or analyst-facing memory dashboard.	Atlas display or disposition is not memory admission, canonization, or truth.	DILIGENCE
Clavis Content Pack Aliases: Clavis	An optional, separately governed, restricted profile or content pack for the sequence-engine ecosystem.	Comparable to a premium or regulated domain pack.	It is not part of the core, not default-installed, and grants no symbolic or ritual authority.	RESEARCH
Coherence Event Signatures (CES) Aliases: CES	Typed features or signatures attached to events so patterns can be compared and traced.	Comparable to event feature engineering or semantic telemetry.	A signature is a diagnostic feature, not an established scientific law.	RESEARCH
CoherenceLattice	The canonical analysis, contract, metric, state, and experiment substrate for the Triadic Brain.	Comparable to an evidence synthesis, workflow-state, and analytical kernel.	It characterizes and coordinates; it does not certify truth, compliance, or release.	CORE
Control Package	A versioned bundle of policies, schemas, documentation, fixtures, tests, and bounded runtime configuration.	Comparable to an installable governance module or compliance-control pack.	Availability, installation, entitlement, and control effectiveness are separate states.	CORE
Governed Sequence Engine (GSE) Aliases: GSE	The neutral kernel for validating, compiling, planning, rendering, closing, and receipting governed sequences.	Comparable to a workflow compiler or policy-aware orchestration engine.	A SequenceSpec cannot authorize itself; runtime remains gated by AEGIS, Sophia, human authority, and phase maturity.	DILIGENCE
Human authority	The human role or body that may approve scoped actions, release, promotion, retention, or canonization.	Comparable to accountable business owner, reviewer, or control owner.	A model, metric, receipt, role label, or AI agent cannot impersonate this authority.	CORE
Local Review Runtime	The bounded local workflow that turns selected sources and candidate outputs into inspectable review artifacts.	Comparable to a local evidence-review and audit-assist product surface.	Local review output remains diagnostic and review-supporting, not final authority.	DILIGENCE
Provenance Memory Reservoir (PMR) Aliases: PMR	The governed memory substrate for provenance-bearing artifacts, retention posture, replay eligibility, and revocation.	Comparable to a policy-aware knowledge store and artifact ledger.	PMR is not automatic memory, model training data, Atlas canon, or a token economy.	CORE
Publisher	The rendering and publication-facing layer for bounded user or public outputs.	Comparable to a reporting, export, and presentation surface.	Publication does not convert claims into accepted evidence or product release.	CORE
Retrosynthesis	A bounded method for working backward from observed patterns or perturbations to candidate hidden structures.	Comparable to root-cause exploration or reverse hypothesis generation.	It generates candidates, not proven causes or truths.	RESEARCH
Sonya	The model, tool, sensor, and modality mediation layer that packages bounded candidate outputs and declares capabilities.	Comparable to a provider gateway, adapter layer, or model broker.	Sonya does not decide whether her own candidate is admissible or final.	CORE
Sophia	The bounded governance and review-routing layer that can clarify, hold, reject, quarantine, escalate, or request evidence.	Comparable to policy reasoning, risk review, and governance workflow orchestration.	Sophia may narrow AEGIS-eligible routes but may not broaden them or execute actions.	CORE
Telemetry Aperture Controller (TAC) Aliases: TAC	A policy for what telemetry is visible, collected, retained, or withheld in a given context.	Comparable to observability sampling, privacy filters, and data minimization controls.	More telemetry is not always safer; aperture must respect privacy and purpose.	DILIGENCE
Thought-Exchange Layer (TEL) Aliases: TEL	The event and graph representation of reasoning, transformations, decisions, and transitions.	Comparable to observability events, workflow lineage, or a cognitive event bus.	TEL is not consciousness proof, evidence by itself, or action authority.	DILIGENCE
Triadic Call Engine	A proposed UVLM product composition built over the neutral Governed Sequence Engine.	Comparable to a packaged governed workflow product.	It is not a new brain organ and is not synonymous with the neutral kernel.	RESEARCH
Universal Control Codex (UCC) Aliases: UCC	A control grammar for how reasoning and governance steps are constrained and reported.	Comparable to policy-as-code or a control framework.	UCC alignment is not regulatory compliance by itself.	DILIGENCE

Governance, authority, and safety

Term	Investor-ready definition	Association / commercial relevance	Boundary	Tier
Accepted-evidence authority	The explicit authority to mark material as accepted evidence for a declared process.	Comparable to evidence admission or record acceptance.	Grounding, provenance, or publication visibility alone does not grant it.	DILIGENCE
Action authority	The explicit right to cause an external, side-effecting, destructive, financial, or privileged operation.	Equivalent to transaction authorization.	Model output, governance review, or a metric cannot grant it by themselves.	CORE
Action Firewall	The AEGIS stage that evaluates whether a proposed operation may proceed toward preflight.	Comparable to a transaction firewall or privileged-action policy gate.	A permitted action category still requires operation-specific authority and execution receipts.	CORE
Admission	The first decision about whether a request or artifact may enter a governed workflow.	Analogous to zero-trust admission control.	Admission means process eligibility, not truth or product approval.	DILIGENCE
Authority ceiling	The maximum authority a component, role, or receipt may exercise.	Comparable to scoped permissions or delegated authority limits.	No component may infer a higher ceiling from its own output.	CORE
Consent	A recorded, scoped, revocable authorization for a declared use.	Commercially relevant to privacy, user trust, and regulated workflows.	Consent is purpose- and operation-specific; it is not perpetual or transferable by default.	CORE
Contestability	The ability to question, appeal, correct, supersede, revoke, or review a decision.	Important for trust, regulated use, and human oversight.	Contestability does not guarantee the preferred outcome; it guarantees a governed path.	CORE
Decision vocabulary	The controlled set of outcomes such as allow, allow with controls, hold for human review, reject fail closed, and alarm for elevated review.	Makes governance observable and machine-testable.	These outcomes describe routing posture; none certifies truth or legal status.	DILIGENCE
Fail closed	When required evidence, authorization, scope, or policy is missing, the system blocks or holds rather than guessing.	A standard safety posture in secure systems.	Fail-closed behavior may reduce convenience; it should not be “fixed” by weakening controls.	CORE
Final-answer authority	The explicit authority to designate an output as the released answer to the user or downstream consumer.	Separates drafts and candidates from official outputs.	A candidate, receipt, or review summary does not automatically possess it.	DILIGENCE
Governance admissibility	A categorical posture indicating whether a case satisfies required governance gates for its bounded scope.	Comparable to policy eligibility.	It is not legal approval, compliance certification, or a scalar score.	DILIGENCE
Grounding	Binding a request or claim to explicit source evidence and source limitations.	Comparable to retrieval grounding and evidence-backed generation.	Grounded does not mean true; source quality and interpretation remain reviewable.	CORE
Human Authority Receipt	A planned credential-bearing record of a human actor’s role, scope, case, operation, expiry, revocation, and confirmation.	Comparable to a digitally scoped approval record.	An AI system may validate such a receipt but may not create the human approval itself.	DILIGENCE
Hyperreal drift	When documents, names, dashboards, and simulations begin to feel more real or mature than the validated system they describe.	A project-risk term for narrative maturity outrunning operational reality.	It is not a claim that all abstraction is harmful; the control is evidence-linked reality checks.	DILIGENCE
Instruction quarantine	Separating instructions embedded in external content from authorized system or user instructions.	A core defense against indirect prompt injection.	Quarantined content may remain evidence but cannot control tools or policy.	CORE
Interaction Autonomy Risk	A proposed assessment of system behaviors that could pressure, isolate, manipulate, or reduce a user’s ability to disengage.	Comparable to human-AI interaction risk and coercion-resistance review.	It evaluates system conduct, not user pathology, spirituality, worth, or diagnosis.	RESEARCH
Least privilege	Each component receives only the minimum capabilities needed for its scoped task.	A core security and governance principle.	Capability availability does not imply permission to use it in every case.	CORE
Local Runtime Preflight	A final local eligibility check before a bounded operation.	Comparable to pre-execution policy and authorization checks.	Preflight allowed does not mean the operation was performed.	CORE
Model Candidate Gate	The AEGIS stage that determines whether a model-produced candidate may proceed to review.	Comparable to output policy validation.	Candidate eligibility is not final-answer or action authority.	DILIGENCE
Monotonic narrowing	Downstream governance may reduce eligible routes but may not expand permissions beyond upstream AEGIS and human scope.	Comparable to non-escalating policy composition.	A downstream component cannot use “good judgment” to bypass a failed upstream gate.	DILIGENCE
Negative control	A test fixture designed to fail, hold, or expose a known error mode.	Comparable to adversarial and falsification testing.	Passing negative controls does not prove general safety or scientific validity.	DILIGENCE
Receipt Chain Export	A bounded export of linked decision and artifact receipts for review or reproducibility.	Comparable to an audit bundle or chain-of-custody export.	Export does not certify truth, compliance, or accepted evidence.	DILIGENCE
Review route	An eligible path for clarification, evidence review, escalation, correction, or human decision.	Comparable to workflow routing in enterprise governance.	A review route is not an action permission.	CORE
Revocation	A governed withdrawal of consent, authority, access, memory use, or prior disposition.	Commercially important for privacy, lifecycle control, and incident response.	Revocation must propagate; it is not a display-only label.	CORE
Role-based human gate	A requirement that a qualified role review a specific request, plan, or operation.	Comparable to separation-of-duties approval workflows.	Roles must not be hard-coded to a person in universal policy.	CORE
Runaway bloat	Growth in phases, schemas, registries, tests, or concepts faster than validated user capability.	An execution and capital-efficiency risk.	Not all complexity is waste; the test is whether each artifact closes a named gap.	CORE
Separation of duties	Different sensitive responsibilities are held by different roles or components.	A familiar enterprise governance and fraud-control principle.	No organ should generate, approve, execute, and certify the same high-impact action.	CORE
Source scope	The exact sources, files, excerpts, connectors, or data ranges the user authorized for use.	Comparable to data-access scope and purpose limitation.	Absence or ambiguity fails closed; directory discovery is not implied.	DILIGENCE
Stop-the-line condition	A predefined condition that pauses a phase or release for review.	Comparable to a quality or safety escalation trigger.	Only evidence-backed triggers should stop the project; personal uncertainty pauses the reviewer, not automatically the phase.	CORE
UI Preflight Status Surface	A presentation-only view of preflight status, reason codes, receipts, and next steps.	Comparable to a control-status dashboard.	The UI is not the enforcement endpoint.	DILIGENCE

Evidence, provenance, contracts, and data

Term	Investor-ready definition	Association / commercial relevance	Boundary	Tier
AI Work Event / Receipt	The canonical spine for recording what AI-related work was requested, decided, performed, or refused.	Comparable to a transaction/event ledger for AI work.	A receipt records an event or decision; it does not certify correctness or effect.	CORE
Artifact	Any governed file, packet, report, receipt, model output, schema, or derived object.	Comparable to a versioned work product.	Artifact existence is not validation, acceptance, or release.	CORE
Artifact reference Aliases: artifact_ref	A stable pointer to an artifact, usually paired with identity and hash metadata.	Comparable to a content-addressed or registry reference.	A reference is not permission to load, execute, publish, or retain the artifact.	DILIGENCE
Candidate packet	A bounded model or system output packaged for review, with identity, provenance, uncertainty, and boundaries.	Comparable to a structured proposal or draft artifact.	A candidate is not a conclusion, action, memory, or final answer.	CORE
Canonical JSON	A stable serialization used so equivalent data produces the same hash.	Supports deterministic receipts, signatures, and replay.	Canonical encoding does not validate the meaning of the content.	DILIGENCE
Canonical owner	The repository artifact or phase that has authority to define a term, schema, formula, or contract.	Prevents duplicate standards and semantic fragmentation.	Canonical ownership is technical stewardship, not product or truth authority.	CORE
Claim	A proposition asserted by a source, model, reviewer, or publication artifact.	The basic unit of evidence review.	Claims must remain distinguishable from source IDs, filenames, sentiments, and actions.	CORE
Content reference Aliases: content_ref	A reference to the content associated with a candidate part or artifact.	Supports content-addressing and deduplication.	A fabricated or null content ref must not bypass required evidence.	DILIGENCE
Contract	A versioned agreement about inputs, outputs, assumptions, guarantees, and boundaries between components.	Comparable to interface contracts and service-level semantics.	A contract describes obligations; it does not prove they were fulfilled in a given run.	CORE
Decision receipt	A record of a governance or review disposition and the evidence, reasons, and authority scope behind it.	Comparable to a decision log.	A decision receipt does not imply the action was performed.	DILIGENCE
Deterministic replay	Reconstructing a prior state or output from the same governed inputs, versions, and event history.	Supports reproducibility, incident analysis, and trust.	Replay equivalence does not prove the original claim was true.	CORE
Event sourcing	A state-management pattern in which changes are recorded as events and current state is reconstructed from them.	Useful for CognitiveRun persistence, audit, and recovery.	Events require versioning, ordering, idempotency, and privacy controls.	DILIGENCE
Evidence packet	A structured set of source evidence, citations, limits, and review status.	Comparable to a due-diligence packet.	Evidence packets support review; they do not become accepted evidence automatically.	CORE
Failure receipt	A receipt emitted when a request is held, rejected, or alarmed, preserving reason codes and non-actions.	Important for fail-closed transparency and postmortem review.	A failure receipt is not itself proof of harm or legal violation.	CORE
GroundingBundle	A deterministic bundle of source manifests, segments, hashes, and evidence context used to ground a request or candidate.	Comparable to an evidence dossier or retrieval bundle.	GroundingBundle presence is not source quality certification.	DILIGENCE
Idempotence	Applying the same governed operation more than once produces the same result or no additional side effect.	Important for safe retries and distributed workflows.	Idempotence must be tested; deterministic text alone does not guarantee it.	DILIGENCE
Lineage	The linked sequence of source, derivative, event, policy, and receipt relationships.	Comparable to dependency and event lineage.	Lineage must not be confused with identity continuity or authority.	DILIGENCE
Manifest	A machine-readable inventory of files, hashes, roles, versions, and references.	Comparable to a bill of materials or release manifest.	Manifest inclusion is not approval or product release.	CORE
Profile	A scoped configuration or policy specialization of a broader contract.	Comparable to an implementation profile or domain configuration.	A profile must not silently redefine the kernel or universal semantics.	DILIGENCE
Provenance	The record of origin, transformation, custody, and responsibility for an artifact or claim.	Comparable to data lineage and chain of custody.	Provenance preserves history; it does not prove truth.	CORE
Receipt	A structured record of a request, decision, transformation, operation, or refusal.	Comparable to an immutable transaction receipt.	Receipts support auditability but do not certify truth, compliance, or effect.	CORE
Registry	A canonical index of schemas, phases, artifacts, formulas, packages, or capabilities.	Comparable to a service catalog or metadata registry.	Competing registries create semantic drift; canonical ownership matters.	CORE
Repair packet	A derivative artifact that corrects a contract or representation defect while preserving source immutability.	Comparable to a migration output or normalized derivative.	A repair packet is not a rewrite of history or proof that the source was valid.	DILIGENCE
RequestEnvelope	The canonical machine-readable request boundary carrying scope, context, source refs, policy, and non-authority metadata.	Comparable to a typed API request plus governance context.	It describes a request; it does not authorize itself.	DILIGENCE
Schema	A formal definition of allowed fields, types, required values, and structures.	Comparable to an API or data-contract schema.	Schema validity is not semantic validity, scientific validity, or authority.	CORE
SHA-256 hash Aliases: hash	A cryptographic digest used to identify exact bytes and detect change.	Comparable to content fingerprints and supply-chain integrity checks.	A hash is not encryption, source quality, authenticity without trust anchors, or truth.	CORE
Source ID versus Claim ID	Source IDs identify where material came from; claim IDs identify propositions made about or from that material.	Prevents database and review semantics from conflating documents with findings.	A source filename, sentinel, or source ID must not appear in claim-result arrays.	DILIGENCE
Source span	The exact excerpt or byte/character range supporting a claim.	Comparable to pinpoint citation or record-location metadata.	A span must retain source identity and context; extraction can introduce loss.	DILIGENCE
State capsule	A read-only snapshot of live Git, remote, PR, phase, and authority state used for phaselock continuity.	Comparable to an operational handoff packet.	A capsule describes state; it does not authorize merge, push, or phase transition.	CORE
Trace	A distributed execution or request-correlation record.	Comparable to application tracing or observability spans.	Trace IDs are not source identity, persona identity, evidence, or continuity proof.	DILIGENCE

Cognition, planning, learning, and memory

Term	Investor-ready definition	Association / commercial relevance	Boundary	Tier
Ablation	A controlled experiment that removes one component to measure its contribution.	Standard investor-diligence evidence for whether architecture adds value.	Ablations should be preregistered and interpreted within scope.	DILIGENCE
Action-observation loop	The closed loop from plan step to AEGIS preflight, bounded execution, observation capture, and state update.	The missing core of a fully functional cognitive runtime.	Every side effect remains separately authorized and receipted.	CORE
Belief ledger	A proposed ledger of propositions, supporting evidence, counterevidence, uncertainty, contradictions, and revision history.	Comparable to an evidence-aware world model.	It tracks epistemic posture; it does not certify truth.	CORE
Capability profile	A versioned declaration of what a model or tool supports, including modalities, structured output, privacy, cost, and network needs.	Enables provider-neutral routing and procurement flexibility.	Declared capability is not permission or quality proof.	CORE
ClosureProtocol	A required terminal procedure for a rendered, refused, held, or incomplete sequence.	Comparable to a safe shutdown or bounded completion protocol.	Closure is not symbolic sealing and does not prove psychological effect.	DILIGENCE
Cognitive workspace	The active state for goals, plans, evidence, beliefs, budgets, capabilities, and unresolved questions.	Comparable to working state in a durable agent runtime.	Workspace contents remain scoped, expiring, and governed.	DILIGENCE
CognitiveRun	The proposed persistent state object for one bounded cognitive task across observation, planning, action, review, memory, and termination.	Comparable to a durable workflow instance or agent run.	A CognitiveRun is not a conscious subject or autonomous legal actor.	CORE
Compiled Cognitive Plan	An immutable, versioned plan derived from a candidate after policy and contract checks.	Comparable to a compiled workflow artifact.	Compilation does not grant action authority.	DILIGENCE
Contradiction graph	A graph linking claims and evidence that cannot all be accepted under the current assumptions.	Supports revision, dissent, and replanning.	Contradiction may arise from scope, time, or definition differences; it is not automatically error.	DILIGENCE
Cross-model dissent	Preserving disagreement between models instead of collapsing it into a vote.	Supports robust review and reduces false consensus.	Dissent is not automatically superior; it must remain evidence-linked.	CORE
Episodic memory	Records of what happened in prior governed runs.	Comparable to event history or case memory.	Prior events are not automatically reusable evidence.	DILIGENCE
GoalSpec	A typed statement of the user’s goal, constraints, success criteria, and stop conditions.	Comparable to a job specification.	A goal does not grant tools, memory, or external-action authority.	DILIGENCE
Memory hierarchy	The separation of working, episodic, semantic, procedural, and prospective memory roles.	Comparable to layered memory architecture in cognitive systems.	No memory class bypasses PMR, consent, privacy, expiry, or revocation.	DILIGENCE
Observation packet	A typed record of what a tool, model, sensor, or source returned after a bounded step.	Comparable to a tool result or environment observation.	Observations require provenance and may still be incomplete or wrong.	CORE
Plan candidate	A proposed sequence of steps with preconditions, postconditions, risks, budgets, and closure.	Comparable to a workflow draft.	It remains reviewable and nonexecuting until compiled and preflighted.	DILIGENCE
Planner / decomposer	The component that proposes or repairs a sequence of subgoals and expected observations.	Comparable to a planning engine in agent systems.	The planner proposes; it does not authorize or execute.	CORE
Procedural memory	Reusable workflows or skills that have passed replay, controls, and human promotion.	Comparable to a validated automation recipe.	A successful plan is only a skill candidate until promoted.	DILIGENCE
Prospective memory	Scoped reminders or deferred goals for future action.	Comparable to a governed task queue.	It must expire, remain revocable, and not self-trigger outside policy.	DILIGENCE
Raw baseline	The same underlying model used without the Triadic governance and cognition layers.	Needed to prove incremental value.	Baseline comparisons must control evidence, tools, budgets, and human work.	CORE
Recall Decision Receipt	A proposed record showing whether memory retrieval was allowed, held, or rejected after hard gates and ranking.	Comparable to a read-authorization receipt.	Relevance cannot override consent, purpose, privacy, rights, expiry, or revocation.	DILIGENCE
Retention candidate	A proposal that an artifact may be retained, pending separate PMR review.	Separates useful output from automatic memory write.	Candidate emission does not perform a memory write.	CORE
Selective recall	Retrieving only eligible memories after hard authorization gates, then ranking them for relevance.	Supports privacy-preserving and context-aware memory.	Selection is not evidence acceptance or identity continuity.	RESEARCH
Semantic memory	Reviewed concepts, claims, and relationships retained with evidence and status.	Comparable to a governed knowledge base.	Semantic memory remains revisable and noncanonical unless authorized.	DILIGENCE
SequencePolicyProfile	The policy rules governing a sequence’s admissible uses, steps, and review requirements.	Comparable to external policy configuration.	Policy remains external to the sequence specification.	DILIGENCE
SequenceSpec	A design-time description of what a governed sequence could render or request.	Comparable to a declarative workflow specification.	A SequenceSpec cannot authorize itself.	DILIGENCE
Skill candidate	A plan or workflow proposed for future procedural reuse after replay and review.	Comparable to a learned automation candidate.	It is not automatically installed, activated, or trusted.	DILIGENCE
Subgoal	A bounded intermediate objective derived from a GoalSpec.	Supports hierarchical planning and measurable progress.	Subgoals remain subordinate to the original scope and authority ceiling.	DILIGENCE
Termination controller	The logic that stops a cognitive run on success, refusal, clarification, review, budget, cancellation, or failure.	Controls cost, safety, and runaway loops.	Termination must be explicit; silence or process death is not a valid completion state.	CORE
Working memory	Short-lived state used inside the active CognitiveRun.	Comparable to a scratchpad or workflow context.	It should expire or be transformed under explicit policy.	DILIGENCE

Metrics, diagnostics, and scientific vocabulary

Term	Investor-ready definition	Association / commercial relevance	Boundary	Tier
36-cell Mutual Containment matrix	The cross-product of six domain dimensions and six relational-obligation families.	Provides comprehensive interface-coverage design.	Matrix completeness is design coverage, not empirical validation.	DILIGENCE
Agency Profile Vector (APV) Aliases: APV	A multi-component profile of bounded operational characteristics such as determinism, reasons responsiveness, goal regulation, containment, and corrigibility.	Comparable to a model-behavior profile rather than a single score.	Not free-will detection, personhood, moral worth, legal capacity, or a general human score.	RESEARCH
Applicability	Whether a metric, control, or dimension belongs to the declared observation scope.	Prevents meaningless scores and false comparisons.	Not applicable is excluded; it must not be silently treated as zero.	CORE
Asymmetric harm protection	The obligation for more capable or authoritative parties to carry stronger protections toward more vulnerable parties.	Comparable to power-aware risk controls.	Asymmetry does not imply malicious intent or legal liability.	RESEARCH
Boundary integrity	The obligation to preserve declared interface, identity, and scope boundaries.	Comparable to interface isolation and integrity controls.	It does not require components to be isolated from all exchange.	DILIGENCE
Calibration profile	The versioned population, weights, thresholds, and methods used to interpret a metric.	Enables comparability and controlled product claims.	Results under incompatible profiles must not be directly compared.	DILIGENCE
Coherence Energy (E) Aliases: E	A research metric family for internally consistent or coupled structure; current implementations may use narrower local-review proxies.	Investor association: signal quality or internal consistency.	Not a validated universal measure of empathy, truth, cognition, or physical energy.	RESEARCH
Confidence	How strongly the evidence or evaluation supports the reported result.	Comparable to evidence quality or assessment confidence.	Confidence is separate from the metric value and from probability unless calibrated.	CORE
Consent and scope	The obligation to preserve authorization, purpose, and data-use limits across a boundary.	Comparable to privacy and access-control obligations.	Consent is contextual and revocable.	DILIGENCE
Containment breach	A categorical finding that a declared boundary was actually crossed, bypassed, overwritten, impersonated, or exceeded.	Comparable to a control breach or policy violation event.	Does not by itself prove malicious intent, legal violation, or audit failure.	CORE
Containment pressure	A bounded measure of how strongly a boundary is being challenged.	Comparable to threat or policy-pressure telemetry.	High pressure can coexist with successful containment and no breach.	CORE
Derived diagnostic	A metric computed from other components for interpretation or comparison.	Comparable to a composite KPI.	Derived diagnostics remain visible with their components and do not create authority.	DILIGENCE
Effective source count	A diagnostic estimate of how many meaningfully independent source families contribute.	Useful for anti-Sybil and consensus-quality analysis.	It is not a truth count and remains sensitive to dependency assumptions.	RESEARCH
Entropy Drift (ΔS) Aliases: Delta S, ΔS	A diagnostic of increasing or decreasing disorder, instability, fragmentation, or unresolved contradiction over time.	Comparable to drift and instability monitoring.	The exact operational definition is context-specific and must be versioned.	RESEARCH
Ethical Symmetry (Eₛ) Aliases: Es, E_s	A research family for how benefits, harms, burdens, and protections are distributed.	Comparable to fairness and affected-party analysis.	Current proxies may measure review protections rather than full social impact.	RESEARCH
G_mu Aliases: Gμ	The categorical Mutual Containment result: true for pass, false for fail, and null for hold or not-applicable.	Separates hard status from soft diagnostics.	It is not a theorem, compliance result, or action authorization.	DILIGENCE
Goodhart risk	When optimizing a metric causes the metric to stop reflecting the intended goal.	Relevant to score gaming and self-certification.	Mitigate with raw observations, hard gates, ablations, and independent review.	DILIGENCE
Hard gate	A noncompensatory requirement that must pass before a route or result can proceed.	Comparable to a mandatory control.	No average or high scalar may override a failed critical hard gate.	CORE
Hold for context	Processing pauses because required evidence, direction, or context is missing or indeterminate.	Comparable to a review queue or clarification state.	A hold is not a pass, failure, or project stop unless separately escalated.	CORE
Lineage integrity	A cross-cutting gate checking event, source, policy, consent, revocation, replay, and receipt continuity.	Comparable to temporal chain-of-custody integrity.	It is not a seventh Mutual Containment dimension or independent invariant.	DILIGENCE
Metric proxy	A narrower operational measurement standing in for a broader research construct.	Allows early product instrumentation while preserving scientific restraint.	A proxy must be named and scoped; it must not masquerade as the full canonical construct.	DILIGENCE
Mutual Containment (μ / M_c) Aliases: μ, M_c, mutual_containment	The candidate engineering invariant for reciprocal boundary preservation across source, model, memory, action, crossmodal, and governance domains.	Comparable to interoperability safety and non-absorption across interfaces.	A design candidate, not a physical law, consciousness measure, compliance score, or route authority.	CORE
Not applicable Aliases: N/A	A dimension or control is outside the declared scope.	Supports clean product and risk profiles.	Not applicable is not missing, zero, or failure.	CORE
Phase-Lock metric (Λ) Aliases: Λ	A research family for synchronization, criticality, or boundary pressure; current implementations must state which semantic profile is used.	Comparable to coordination or threshold pressure.	Do not confuse with the project-management process “phaselock.”	RESEARCH
Primitive invariant	A metric family represented independently in the engineering roster.	Investor association: foundational research variable.	“Independent” in the roster does not prove statistical independence or physical invariance.	RESEARCH
Psi (Ψ) Aliases: Psi, Ψ	A derived coherence diagnostic historically expressed from other metric components such as E and T.	Investor association: composite diagnostic.	Ψ is not an independent invariant and must not become a universal “truth” or safety score.	RESEARCH
Relational obligation	A reciprocal duty evaluated across a Mutual Containment boundary, such as consent, contestability, or least privilege.	Comparable to interface control objectives.	Obligations enrich the contract but do not create new invariants.	DILIGENCE
Reversibility and exit	The obligation to preserve rollback, disengagement, and safe termination where practical.	Supports user control and operational resilience.	Irreversible operations require stricter gates.	DILIGENCE
Source-family independence	Adjusting apparent consensus for shared models, prompts, corpora, editors, or retrieval roots.	Comparable to anti-correlation or evidence-independence analysis.	Many agreeing outputs may still be one correlated source family.	RESEARCH
TEVV Aliases: Testing, Evaluation, Verification, and Validation	Testing, Evaluation, Verification, and Validation.	A standard assurance umbrella for technical and human evidence.	Verification of requirements is not validation of user value or scientific meaning.	CORE
Total Action Functional (TAF) Aliases: TAF	A research diagnostic for the cost or burden of action across informational, coherence, and runtime terms.	Comparable to a composite operational effort or friction measure.	Current runtime proxies are not the full canonical physical/informational theory.	RESEARCH
Transparency (T) Aliases: T	A diagnostic of source visibility, inspectability, reproducibility, and reasoning exposure.	Comparable to auditability or explainability infrastructure.	Transparency does not guarantee correctness or fairness.	CORE
Uncertainty	The known limits, interval, ambiguity, or lack of calibration around a result.	Essential for investor diligence and regulated use.	Low uncertainty does not imply truth; uncalibrated outputs must avoid false precision.	CORE
Unknown	A required value cannot be determined from available evidence.	A first-class state supporting honest uncertainty.	Unknown must not be silently converted to pass, fail, or zero.	CORE
Veto and contestability	The obligation to preserve the ability to question, refuse, appeal, or stop an interaction.	Relevant to user trust and accountable AI.	A veto route must be operational, not merely documented.	DILIGENCE
μ_balance	A geometric-balance diagnostic across applicable Mutual Containment components.	Shows whether containment is balanced rather than concentrated.	It cannot override a failed hard gate.	DILIGENCE
μ_diagnostic	A combined diagnostic using weakest-link and balance information.	Useful for trend and comparative analysis.	Diagnostic only; no route, action, memory, release, or compliance authority.	DILIGENCE
μ_floor	The weakest applicable Mutual Containment component or edge.	Investor association: weakest-link diagnostic.	A low value does not prove illegality; a high value does not prove safety.	DILIGENCE

Product, maturity, development, and operations

Term	Investor-ready definition	Association / commercial relevance	Boundary	Tier
Accepted Phase Matrix	A machine-readable list of phases accepted into a declared project state.	Comparable to a controlled configuration baseline.	Accepted phase status does not imply runtime, product, or scientific validation unless stated.	DILIGENCE
Active lane	The currently authorized sequence of work for a bounded dependency or product area.	Comparable to a workstream.	Multiple lanes may exist, but only one canonical patch should change a given owner at a time.	DILIGENCE
Artifact Index	A searchable inventory of artifacts and their roles, refs, and provenance.	Improves due diligence and operational traceability.	An index is descriptive, not authoritative evidence acceptance.	DILIGENCE
Bounded pilot	A pilot with explicit user, data, capability, duration, rollback, and authority limits.	The preferred transition from prototype to reality testing.	Results remain context-specific.	CORE
Canonical patch	The single bounded implementation specification governing one Codex iteration or PR.	Comparable to an approved change request and acceptance contract.	A summary, memo, or consultant report does not replace it while Codex is running.	CORE
Claim Boundary Index	A registry of allowed, blocked, and bounded claims associated with phases and artifacts.	Reduces marketing, legal, and investor overclaim risk.	The index supports claim discipline; it is not legal advice.	CORE
Codex iteration	A bounded implementation run that executes one canonical patch on a named branch.	Comparable to an AI-assisted development sprint.	Codex does not choose the roadmap, merge, push, or grant release authority.	CORE
compileall	A Python syntax/import-bytecode check used to detect syntax errors across source and tests.	A lightweight engineering quality gate.	Passing compileall is not functional or scientific validation.	DILIGENCE
Compliance-ready report	A report structured to support compliance evidence review and mapping.	Commercially relevant to regulated enterprises.	“Compliance-ready” means evidence-supporting, not compliant, certified, or legally sufficient.	CORE
Control Package Manifest	The versioned declaration of a control package’s contents, dependencies, policies, and boundaries.	Comparable to a package manifest or software bill of materials.	A manifest does not prove installability or effectiveness.	DILIGENCE
Control Package Registry	The canonical index of available control-package definitions and versions.	Supports modular product packaging and governance.	Registry presence is not customer entitlement or activation.	DILIGENCE
Design-only Aliases: design_only	A phase or artifact defines architecture, doctrine, or contracts without implementing runtime behavior.	Important for separating IP development from product claims.	Design-only does not mean validated, installed, or available to customers.	CORE
Evidence Review Pack	A bundled set of evidence, claims, review prompts, metrics, and receipts for human examination.	Comparable to a diligence or case-review packet.	A review pack is not accepted evidence or audit certification.	CORE
Experiment Suite Registry	The canonical index of experiments, phases, tests, and expected evidence.	Supports reproducibility and roadmap control.	Registry inclusion is not successful execution.	DILIGENCE
General availability Aliases: GA	A product is broadly offered with support, security, operations, and commercial commitments.	Investor shorthand for market launch maturity.	The current research and prototype artifacts should not be described this way unless formally approved.	CORE
git diff --check	A Git hygiene check for whitespace errors and conflict markers.	A simple repository-quality control.	It does not prove semantic correctness or a clean release.	DILIGENCE
Guided Review	A user interface or workflow that helps a reviewer inspect evidence and record decisions while preserving review provenance.	Supports adoption by nontechnical users.	Guidance must not fabricate findings or convert sentinels into claims.	CORE
Human Review Closure	The final structured summary of what a human reviewed, what was assessed, what was not assessed, and what remains unresolved.	Comparable to a review sign-off record.	Not assessed must remain distinct from assessed-none-found.	DILIGENCE
Live bounded local Aliases: live_bounded_local	A locally running, tightly scoped prototype with explicit non-authority and environment limits.	A useful pre-product maturity stage.	Not general availability, production hardening, or broad customer readiness.	CORE
Minimal Viable Receipt (MVR) Aliases: MVR	The smallest useful, human-readable, machine-checkable receipt proving what a bounded AI workflow did or did not do.	A potential product wedge for auditable AI work.	Minimal does not mean incomplete authority boundaries or weak provenance.	CORE
Phase ID	A stable identifier for a bounded project change, experiment, repair, or design decision.	Creates investable execution traceability.	A phase name is not proof that the capability exists.	CORE
Phaselock	The operating discipline of keeping repository state, phase ID, canonical patch, authority, validation, and publication in sync.	Comparable to controlled change management and release-train discipline.	Do not confuse with the Λ metric. A reviewer’s pause does not automatically stop the project.	CORE
Pilot	A controlled trial used to gather evidence in a bounded environment.	Investor association: de-risking and customer validation.	A pilot is not general availability or proof of scalability.	CORE
Product Evidence Spine	The canonical evidence chain connecting product claims, tests, validators, artifacts, maturity, and review.	Comparable to a release evidence ledger.	It supports readiness decisions but cannot self-approve them.	CORE
Product readiness	A human-reviewed judgment that product, support, security, privacy, operations, and claims are ready for the next release stage.	Comparable to go-to-market readiness review.	Readiness is not inferred from test count or documentation volume.	CORE
Product slice	A coherent subset of the system that delivers one bounded customer outcome.	Useful for go-to-market sequencing and anti-bloat discipline.	A slice should close a real workflow, not merely bundle documents.	CORE
Publication-visible	A phase or artifact appears in the public publication repository or dashboard.	Supports transparency and investor diligence.	Visibility is not release, product maturity, or authority.	CORE
Release candidate	A specific build proposed for release after validation and review.	A familiar software commercialization stage.	A release candidate is not released until human authorization and deployment occur.	CORE
Reproducibility Dashboard	A public or internal surface showing accepted phases, artifacts, validation status, and claim boundaries.	Useful for investor technical diligence.	Dashboard status must not exceed the underlying evidence.	CORE
Rollback	Returning to a known prior state after a failed or unsafe change.	A core operational-readiness capability.	Rollback should be tested; a theoretical Git command is not an operational proof.	CORE
Safety branch	A local branch preserving the pre-merge state before risky integration.	Supports rollback and business continuity.	A safety branch does not replace backups or validated release tags.	CORE
Static verified	A static artifact, schema, document, or fixture has passed its declared validation checks.	Comparable to verified documentation or contract assets.	Static verification does not prove runtime or product behavior.	DILIGENCE
Validation evidence packet	The collection of tests, semantic inspections, logs, diffs, hashes, and summaries supporting a phase decision.	Comparable to a change-control evidence pack.	Validation evidence supports human decisions; it does not self-approve them.	CORE

Metaphors, names, and lexical signifiers

Term	Investor-ready definition	Association / commercial relevance	Boundary	Tier
Atlas	A map metaphor for organizing and presenting knowledge, memory posture, and novelty.	Signals navigation and visibility.	A map is not the territory; Atlas is noncanonical presentation.	CORE
Brain	A metaphor for coordinated cognitive functions distributed across specialized components.	Signals integrated cognition and decision support.	Not a biological or conscious-brain claim.	CORE
Bridge	A governed connection between schemas, repositories, phases, or representations.	Comparable to an adapter or compatibility layer.	Bridges must preserve source and authority boundaries rather than flatten them.	CORE
Call	A request or governed sequence invocation, used in the Call Engine concept.	Comparable to a workflow invocation or service call.	Does not imply ritual, summoning, or autonomous action.	RESEARCH
Clavis / key	A metaphor for an optional key, index, or restricted content profile.	Signals access to specialized content or mappings.	Not a master key, default capability, or authority grant.	RESEARCH
Constellation	A metaphor for multiple roles or signals considered together without collapsing into one authority.	Useful for oversight or portfolio views.	Avoid turning oversight roles into anthropomorphic guardian personas.	RESEARCH
Corridor	A metaphor for a promising route through a problem or knowledge space.	Comparable to an opportunity pathway or search corridor.	A corridor is a hypothesis about direction, not evidence of breakthrough.	RESEARCH
Echo	A project persona and metaphor for iterative reflection, synthesis, and continuity.	Signals learning and consultation.	A persona label is not identity continuity, authority, or personhood.	CORE
Gate	A decision point that allows, controls, holds, rejects, or escalates a route.	Comparable to a policy or quality gate.	A gate outcome is scoped and may require later gates.	CORE
Gateway	An ingress or egress boundary for requests, files, models, tools, or receipts.	Comparable to an API gateway.	A gateway must not become a policy bypass.	CORE
Gnosis	A historical/project term for synthesized understanding or a continuity report.	Signals reflective knowledge integration.	Use carefully with investors; it is not privileged truth or spiritual authority.	RESEARCH
Lattice	A structured network of relationships, states, or measurements.	Evokes graph structure and compositional architecture.	The geometric metaphor should not be presented as established physics.	CORE
Light	A brand metaphor for clarity, understanding, and inspectability.	Supports UVLM storytelling.	Clarity language must not become truth-certification language.	CORE
Membrane	A trust boundary that mediates exchange while preserving identity, scope, and authority.	Comparable to an API boundary, policy membrane, or zero-trust interface.	A membrane enables governed exchange; it is not total isolation.	CORE
Organ	A bounded component with a distinct responsibility, interface, and authority ceiling.	Comparable to a service or module in a modular architecture.	No new organ should be added without a proven architectural gap.	CORE
Pattern donor	A concept or structure in one domain that may help solve problems in another.	Comparable to cross-domain transfer or analogical reuse.	A useful analogy can still be scientifically invalid.	RESEARCH
Reservoir	A metaphor for governed storage and potential reuse, used in PMR.	Signals accumulated memory resources.	Storage does not imply retrieval, learning, or authority.	CORE
River	A research metaphor for flows of information or discovery through connected domains.	Useful in visualization and research storytelling.	Not a canonical product or scientific claim unless operationalized.	RESEARCH
Rosetta	A translation and interoperability metaphor for mapping between representations, metrics, or systems.	Signals cross-domain compatibility.	A mapping does not prove semantic equivalence.	RESEARCH
Sequence	An ordered set of governed steps, renderings, reviews, or receipts.	Comparable to a workflow.	Sequence order does not grant step authority.	CORE
Shield / Aegis	A protection metaphor for enforcement, containment, and authority boundaries.	Signals risk reduction and guarded execution.	Does not imply perfect safety or absolute protection.	CORE
Spine	A canonical record or dependency chain connecting events, evidence, and decisions.	Comparable to an event backbone or system of record.	A spine organizes evidence; it does not make every attached artifact authoritative.	CORE
Terrace	A metaphor for stabilized knowledge or a durable plateau in a research landscape.	Signals maturation or consolidation.	Research metaphor, not a maturity label.	RESEARCH
Waters	A research metaphor for dynamic fields of cognitive or informational activity.	Useful for exploratory visualization.	Research-only; avoid in investor claims unless clearly framed as metaphor.	RESEARCH
Wave	A metaphor for phase, propagation, family, or coordinated change.	Used in research and roadmap naming.	Do not imply physical wave behavior without a defined model.	RESEARCH

Acronym and symbol index

Acronym / symbol	Expansion or meaning
UVLM	Ultra Verba, Lux Mentis
AEGIS	Project name for the admission and enforcement membrane
UCC	Universal Control Codex
TEL	Thought-Exchange Layer
CES	Coherence Event Signatures
TAC	Telemetry Aperture Controller
PMR	Provenance Memory Reservoir
GSE	Governed Sequence Engine
MVR	Minimal Viable Receipt
TAF	Total Action Functional
APV	Agency Profile Vector
TEVV	Testing, Evaluation, Verification, and Validation
GUFT	Grand Unified Field Theory/Template of Coherence - research framework, not established physics
TCHES	Thermofractal Capillary Heat-Exchange System - a research fixture, not the universal product
GA	General Availability
N/A	Not Applicable

Glossary governance rules

• This glossary translates project language for investors; exact machine tokens come from canonical registries and schemas.
• A metaphor is not a literal scientific or biological claim.
• A phase name is not evidence that a capability is implemented.
• A receipt, hash, schema pass, dashboard, or publication entry is not truth, compliance, audit, product, or release authority.
• When live status matters, use a fresh phaselock-state capsule and the active canonical patch.
• New jargon should be admitted only when it closes a real semantic gap and does not duplicate a canonical owner.

Final investor doctrine

The investable thesis is not that the system makes models truthful or autonomous. It is that it makes model work more governable, inspectable, revisable, interoperable, and accountable while preserving human authority.