Skip to main content

External Frameworks

Purpose

External frameworks are non-StegVerse systems that may help explain, declare, preserve, compare, review, or reconstruct parts of an admissibility review path.

They are included in the wiki as observatory and crosswalk targets, not as canonical Admissible-Existence formalism sources.

The purpose of this section is to show how external framework outputs can be processed through StegVerse under their own terms while StegVerse preserves the same governed-transition grammar and returns the expected commit-time result.

Observatory Strategy

The External Frameworks section functions as a governance observatory.

An observatory does not rewrite the external framework. It records the framework's own claims, terminology, artifacts, and limits, then separately records how those artifacts map into StegVerse governance primitives.

External framework artifact
|
| native terms preserved
v
StegVerse observation
|
| general governance primitives
v
Commitment Candidate
|
| non-authorizing proposed crossing
v
Standing-Proof-Engine
|
| ALLOW / DENY / FAIL-CLOSED
v
Evaluation report

Core Rule

Historical governance artifacts are useful, but they do not become execution authority.

Declaration != execution authority
Evidence != execution authority
Trace != execution authority
Review != execution authority
Commitment Candidate != execution authority

Execution authority must be reconstructed at the commit boundary from current standing.

Evidence Provenance Standard

External framework pages must distinguish at least five evidence classes:

framework claims
framework implementation evidence
observed behavior
StegVerse analysis
interoperability standing

The governing page is External Framework Evaluation Standard.

The reusable benchmark suite is Runtime Governance Benchmark Suite.

The benchmark rollout tracker is External Framework Benchmark Mapping Rollout.

The broader candidate intake registry is Expanded External Framework Intake.

Candidate-family representation and promotion order are tracked in External Framework Family Coverage.

Candidate promotion is governed by External Framework Intake Promotion Criteria.

The reusable template is External Framework Page Template.

Failure observations should use the External Framework Failure Class Catalog.

The rollout tracker is Evidence Provenance Rollout.

Benchmark Posture

The Runtime Governance Benchmark Suite is framework improvement infrastructure, not a takedown mechanism.

benchmark result != framework invalidity
ALLOW != StegVerse authority
BLOCK != StegVerse certification
ESCALATE != failure
parser limitation != governance conclusion
observed behavior != general compatibility

Each framework may be strongest in a different region. A missed case may identify an implementation gap, a scope boundary, an adapter requirement, a missing policy, or an undeclared assumption.

Terminology Reconciliation

External terminology is governed by the Governed Terminology Reconciliation Rule.

A relationship may be classified only as synonymous, adjacent, new, or unresolved.

Similar wording, intent, implementation, or outcome is insufficient to establish synonymy. No reconciliation may inherit execution authority, admissibility, standing, certification, governance proofs, or the published claims of the external framework.

Evaluation Result Posting

External framework evaluation results are posted from generated compatibility reports, not by manually asserting standing on framework pages.

The current generated results surface is External Framework Evaluation Results.

A generated compatibility result is evidence only. It is not certification, endorsement, formalism adoption, admissibility proof, or execution authority.

Build Strategy Pages

PagePurpose
External Framework Evaluation StandardDefines evidence provenance, claim traceability, comparative fairness, and required page structure for all external-framework pages.
Runtime Governance Benchmark SuiteDefines reusable benchmark cases and boundary classes for testing runtime governance, preparation, commitment, execution, semantic equivalence, unknown trajectories, replay, and reconstruction.
External Framework Benchmark Mapping RolloutTracks how the runtime-governance benchmark dimensions are mapped across all registered external frameworks without treating unmapped dimensions as failures.
Expanded External Framework IntakeTracks additional candidate frameworks and framework families without treating intake candidates as sourced validation evidence.
External Framework Family CoverageMeasures representation across intake classes, defines promotion priority tiers, and identifies future framework-family gaps.
External Framework Intake Promotion CriteriaDefines how source-required candidates become sourced intake, page, mapping, or fixture candidates without creating authority.
External Framework Failure Class CatalogDefines reusable failure classes so observations are classified by failure kind rather than only pass/fail.
External Framework Page TemplateProvides a reusable page structure for new or refactored external-framework entries.
Evidence Provenance RolloutTracks rollout of the provenance standard across all registered external-framework pages.
Governance Observatory ProtocolDefines the neutral page structure and observation rules for external frameworks.
Commit-Time Interoperability ContractDefines the minimal contract for routing external artifacts into a Commitment Candidate and SPE determination.
Governed Terminology Reconciliation RuleDefines how external terms may be classified as synonymous, adjacent, new, or unresolved without inheriting authority.
External Framework Evaluation ResultsGenerated posting surface for compatibility-report results.

Current External Frameworks

The current registered framework table contains sourced pages, page candidates, mapping companions, or fixture-backed records. It is not the full universe of possible external frameworks. Additional candidates are tracked in Expanded External Framework Intake, while family coverage and promotion priority are tracked in External Framework Family Coverage.

FrameworkPrimary RoleCitation StatusRelationship To Admissibility
GLMPre-event boundary declarationsourcedHelps declare claims, non-claims, scope, composition, and interpretation frame before admissibility review.
EVIDEPost-event evidentiary reconstructabilitysourcedHelps preserve and organize evidence after an event so reconstruction and dispute review remain possible.
DecisionAssureTrace integrity and causal-continuity intakeartifact package requiredCandidate target for testing whether supplied traces can support independent authority reconstruction without becoming authority.
MindForgeHistorical governance review evidence intakeartifact package requiredCandidate target for testing whether review evidence remains non-authorizing until SPE reconstructs current standing.
Morrison RuntimeRuntime-governance comparison targetsourced with parameterized boundary case partialCandidate comparison target for runtime outcome, failure-state behavior, semantic-equivalence boundary testing, and preparation-vs-execution boundary testing.
CARE RuntimePublic runtime-governance platform candidateofficial source requiredCandidate crosswalk target for runtime governance, public-platform posture, and consequence-boundary language.
KPTRuntime decision boundary before downstream consequenceofficial source requiredSource-blocked candidate for routing explicit decision-state and trace evidence into independent StegVerse commit-time standing review without inheriting authority.
AARSupervised AI governance and operational forensics candidatesourcedCrosswalk target for observable-system governance and pre-execution cost governance.
ASROIndependent AI governance-state attestationsourcedCrosswalk target for host-side measurement, edge-side witnessing, and independent verification of whether declared governance remained in force.
MITRE ATLASAdversarial AI threat knowledge basesourcedCrosswalk target for tactics, techniques, mitigations, case studies, and threat-informed review posture.
OWASP Top 10 for LLM ApplicationsLLM application risk and vulnerability guidancesourcedCrosswalk target for risk categories, prompt injection, output handling, mitigations, and application security review.
Agent Governance PlaybookAgent-continuation governance playbooksourcedCrosswalk target for whether an agent should still be allowed to continue.
Emergency Stop ConventionAI-agent emergency stop conventionsourcedCrosswalk target for emergency stop conditions and fail-closed behavior.
NIST AI RMFVoluntary AI risk managementsourcedCrosswalk target for risk, trustworthiness, lifecycle review, and evaluation posture.
ISO/IEC 42001AI management system standardsourcedCrosswalk target for organizational AI governance and management-system controls.
EU AI ActLegal/regulatory AI frameworksourcedCrosswalk target for obligations, risk classes, documentation, oversight, and governance duties.
Policy CardsMachine-readable runtime governance artifactsourcedCrosswalk target for runtime policy artifacts, obligations, evidentiary requirements, and allow/deny rules.
Runtime Governance for AI AgentsRuntime path-governance frameworksourcedCrosswalk target for path-dependent runtime checks and proposed next-action review.
Admissible Existence Seed CycleEcosystem seed-cycle status mirrorsourcedMirrors completed AE seed-cycle artifacts while preserving that publication does not create standing or execution authority.
Decision Authority CompatibilityST-004 decision vocabulary compatibilitysourcedCrosswalk target for mapping public wiki decision shorthand into canonical transition-authority values without creating authority.

Three-Part Bridge

GLM asks: What boundary is declared before interpretation?
Admissibility asks: May this transition bind consequence at commit time?
EVIDE asks: What evidence remains after the event?

Expanded Observatory Bridge

GLM -> declaration
EVIDE -> evidence
ASRO -> governance-state attestation
MITRE ATLAS -> adversarial AI threat context
OWASP Top 10 for LLM Applications -> LLM application risk context
DecisionAssure -> trace
MindForge -> review
Morrison Runtime -> pre-execution runtime verdict and audit evidence
KPT -> source-blocked runtime decision state before downstream consequence
Runtime Governance Benchmark Suite -> shared boundary tests
Benchmark Mapping Rollout -> cross-framework applicability tracker
Expanded External Framework Intake -> candidate discovery without validation claims
Framework Family Coverage -> representation and promotion priority
Intake Promotion Criteria -> source-gated promotion path
Decision Authority Compatibility -> decision vocabulary reconciliation
Commitment Candidate -> proposed crossing
SPE -> standing determination
Execution -> consequence binding

Source Policy

Pages marked official source required remain intake records only. They should not be treated as externally sourced until an official website, public canonical artifact, release page, repository, paper, or specification URL is supplied.

Pages marked artifact package required remain intake records until a public source, authorized artifact package, repository, specification, or jointly authorized technical note is supplied.

Pages marked sourced with parameterized boundary case partial have official source references and at least one parameterized observation, but are not full public validation results until raw outputs, timestamps, source versions, and audit payloads are attached.

Pages marked sourced include a public or official source URL on the page itself.

Candidate intake records are not sourced framework pages and should not be used as evidence until canonical sources are attached and the relationship to StegVerse is classified.

Authority Boundary

Admissible-Existence defines canonical formalisms.
Publisher publishes papers and public exposition.
Admissibility Wiki mirrors, relates, crosswalks, and discovers relationships.
External frameworks are mapped here only as observatory and relationship/crosswalk targets.

Non-Claims

External framework inclusion is not equivalence.
External framework inclusion is not formalism adoption.
External framework inclusion does not prove transition admissibility.
External framework inclusion does not grant execution authority.
A source URL is evidence of the external framework artifact, not acceptance of equivalence.
A StegVerse observation does not expand the external framework's published claims.
A parameterized observation is not general compatibility.
Benchmark results are not judgments of framework value.
Benchmark mapping does not make unmapped dimensions failures.
Candidate intake does not make unsourced candidates validation evidence.
Candidate promotion does not create certification or execution authority.
Framework-family coverage does not create validation, ranking, or compatibility claims.
Decision-authority compatibility does not create execution authority or override current standing.