Operational Substrate
The concrete material a claim points to: doctrine, schemas, tests, prompts, reports, hashes, manifests, review boundaries, and maintenance records. A term is not trusted by name alone; it must connect to substrate that can be inspected.
Used in: Start Here, Evidence Architecture
Bounded Evidence Record
A scoped record of what was tested, what happened, what the result means, what it does not mean, and which supporting artifacts are included or hash-bound.
Used in: Home, Evidence Architecture
Canonical Manifest
A deterministic listing of evidence artifacts, file hashes, repo state, sidecars, reports, and non-claims. It gives a receipt a stable object to bind.
Used in: Evidence Architecture
Tamper-Evident Receipt
A receipt that preserves the manifest-bound evidence package from the moment of sealing onward. It does not make the underlying claim true by itself.
Used in: Evidence Architecture
Append-Only Maintenance
A repair discipline where later corrections create new notes, sidecars, commits, receipts, or references instead of silently rewriting the sealed evidence story.
Used in: Evidence Architecture
External Verification
Trust added by independent reproduction, audit, provider attestation, hardware attestation, or other outside review. It exists only when actually supplied, not because a receipt structure exists.
Used in: Evidence Architecture
Replayability
The property that a declared evidence or computation path can be rerun from declared inputs and compared against published outputs. Replayability supports inspection; it is not universal truth.
Used in: Start Here
Admissibility
Sovrient's public-surface shorthand for whether a claim or artifact passes declared evidence gates. It is not courtroom admissibility doctrine unless a separate legal process says so.
Used in: Home, Verification
Anchor
A public reference, transaction, witness, or hash handle that helps locate or verify a published evidence state. An anchor helps preserve a record; it does not prove the underlying claim is true.
Used in: Anchors
Sealed Catalog
The public catalog of sealed evidence artifacts and their verification metadata. It is an inspection surface, not a dashboard-only trust claim.
Used in: Sealed Catalog
Machine State
A compact machine-readable state file for agents and reviewers. It helps software locate current evidence posture, but the underlying artifacts remain the source of truth.
Used in: Machine State
Parametric Trigger
A predefined condition that determines eligibility in a parametric structure without case-by-case loss adjustment. It depends on declared measurements, thresholds, and policy logic.
Used in: FONDEN post
Pre-Catalog Parameter
An event parameter captured before later scientific normalization cycles. It is suited for deterministic trigger-state replay when provenance and timing are explicitly sealed.
Used in: Pre-catalog post
Catalog Parameter
A refined event parameter published in a curated catalog that may change over time. Catalog quality can improve while introducing revision-handling requirements in trigger governance.
Used in: Pre-catalog post
Oracle Risk
The risk that settlement-critical state cannot be independently reproduced under declared scope. Oracle risk rises when systems depend on mutable single-source values without replay controls.
Used in: FONDEN post
Trigger Dispute Surface
The set of ambiguities counterparties may contest when evaluating trigger outcomes. Deterministic replay and sealed evidence reduce this surface.
Used in: FONDEN post
Deterministic Replay
Re-executing a declared computation path from declared inputs to reproduce the same outputs byte-for-byte. Determinism is the minimum technical bar for independent verification.
Used in: Start Here
Fail-Closed
A control posture where missing or mismatched verification artifacts block publication. The system does not downgrade requirements silently when verification fails.
Used in: VOSINT post
Attestation Context
The declared set of inputs, transforms, hashes, signatures, and witness metadata that makes a claim replayable. Context links narrative statements to verifiable artifacts.
Used in: FONDEN post
Non-Repudiation
Cryptographic evidence that a signed state existed at a declared time and cannot be plausibly denied later. Detached signatures and witness roots are typical mechanisms.
Used in: VOSINT post
Provider Drift
Divergence between providers over time in reported event parameters or derived indicators. Drift requires explicit policy handling to avoid implicit trigger inconsistency.
Used in: Market drift lane
Merkle Seal
A root hash committing a set of artifact hashes under deterministic ordering and node rules. It allows inclusion verification without transferring the entire artifact set.
Used in: Verification
GPG Signature
A detached cryptographic signature binding a specific file hash to a signing key. Signature verification confirms authenticity and integrity of published artifacts.
Used in: Verification
DMSS
Deterministic Multi-Source Settlement Standard: a control framework where trigger-critical measurements require multi-source corroboration, deterministic replay, and sealed evidence before admissible publication.
Used in: Start Here
DT0
A strict determinism profile requiring zero unresolved computational variance under declared scope. DT0-compliant runs are either reproducible or fail-closed.
Used in: Verification
DAROC
Deterministic Acquisition, Reconciliation, and Operational Cryptoseal: an execution pattern where acquisition, reconciliation, and cryptographic sealing are bound into one replayable claim-chain.
Used in: Protocol
VOSINT
Verifiable Open-Source Intelligence. Public-source inputs are upgraded to evidentiary quality by verification (V1), replayability (V2), and witnessed non-repudiation (V3).
Used in: VOSINT Framework
NERV Release Evidence Bundle
A bounded package of release evidence, policy criteria, manifests, receipts, and verdict state. On the federal evidence surface, NERV refers to this evidence-bundle pattern rather than a certification or authorization regime.
Used in: Federal AI Evidence
PCBA
Persona Consortium Based Analysis: Sovrient's structured adversarial review method for testing claims across defined reviewer roles. PCBA is Sovrient-specific methodology, not an industry standard or federal certification process.
Used in: Federal AI Evidence, AUTH-REF Doctrine
AIBOM
AI bill of materials: an inventory of AI, data, model, component, and lineage elements relevant to a scoped release. An AIBOM-style inventory supports reviewability; it does not by itself approve use.
Used in: Federal AI Evidence, AI RMF Playbook Twin
AUTH-REF
Authority-bound reference register. Authority-bound means source bytes are bound and checked against a cited authority; it does not mean authority-substituting.
Used in: Federal AI Evidence, AUTH-REF Register
TEVV
Testing, evaluation, validation, and verification artifacts tied to a release boundary. TEVV evidence is useful only when scoped to the system, policy pack, and operating conditions being reviewed.
Used in: Federal AI Evidence
Fail-Closed Gate
A release control that blocks promotion when required evidence is missing, stale, malformed, or mismatched. It preserves the evidence boundary instead of silently downgrading requirements.
Used in: Federal AI Evidence
ANATOP
Agent-Native Topology. Sovrient’s method for turning documents, measurements, and decision structures into deterministic, sealed, bounded machine surfaces that agents can use without hidden interpretation steps.
Public-safe shorthand: ANATOP governs state formation. It is about how raw artifacts become bounded, deterministic, agent-usable state that can be replayed and checked later.
Used in: Document-State Twins, Supporting Methodology
MAS
Sovrient’s governed memory runtime and replayable path-validity engine. MAS handles admissibility, path selection, contradiction handling, and fail-closed emission over declared state.
Used in: Protocol, Maritime AIS
NERV
Net Expected Risk Value. A deterministic decision score computed from declared risk, loss, and verification-cost inputs under fixed ordering and rounding rules.
Used in: Market drift lane
Governance Seed
A bounded early-stage publication artifact proving source possession, manifest discipline, and replay posture without claiming full production maturity.
Used in: Maritime AIS
Evidence-Path Reconstruction
Sovrient’s practice of reconstructing why a machine-supported workflow landed where it did in evidentiary terms: what it saw, what it selected, what it excluded, and what it carried forward. It is not a claim to read inner model consciousness or hidden subjective reasoning.
Used in: Supporting Methodology, Maritime AIS
SSEJ
Saw, Selected, Excluded, Justified: Sovrient shorthand for the supporting methodology invariant historically published as governance reconstruction. A decision record is reconstructable if and only if these four components are present, scope-bound to the same decision, and independently verifiable.
`SSEJ` is operational on Sovrient's public surface today, not only defined conceptually. Current public examples include the live seismic verification surface and the bounded Maritime AIS review workflow. Future-compatible domains include agent decision traces, unmanned mission review, and workflow adjudication where the same scope-bound invariant can be applied.
Roadmap note: SSEJ is compatible with managed-agent runtimes such as Anthropic's Claude Managed Agents, but Sovrient does not currently claim a live public managed-agent decision lane or a shipping Anthropic integration.
Public-safe shorthand: SSEJ governs decision reconstruction. In Sovrient terms, ANATOP governs what becomes trusted state, and SSEJ governs what happens over that state in a reviewable decision record.
Used in: Supporting Methodology, Maritime AIS, Verification
Maritime AIS Governance Surface
A published lane applying Sovrient’s evidence discipline to authoritative open U.S. coastal AIS data. Current scope is a real Hampton Roads slice, a manifest, and bounded non-claims around live/global coverage.
Used in: Maritime AIS