mirai-graph 1.4.0 → 2.1.0

package/profiles/project-management/profile.json

@@ -1,38 +0,0 @@
-{
-  "id": "profile.project_management",
-  "name": "Project Management",
-  "version": "0.1.0",
-  "description": "Profile for modeling goals, tasks, decisions, risks, milestones and evidence in a project.",
-  "allowed_object_kinds": [
-    "system",
-    "goal",
-    "task",
-    "decision",
-    "risk",
-    "milestone",
-    "stakeholder",
-    "evidence",
-    "governance_gate",
-    "project_dna",
-    "stage",
-    "batch",
-    "requirement",
-    "method",
-    "lesson",
-    "regression_case"
-  ],
-  "allowed_relation_types": [
-    "depends_on",
-    "blocks",
-    "supports",
-    "evidences",
-    "belongs_to",
-    "governs",
-    "related_to"
-  ],
-  "default_readiness": "draft",
-  "governance_gates": [
-    "human_review_before_canonical_write",
-    "owner_approval_before_action"
-  ]
-}

package/profiles/skill-runtime/README.md

@@ -1,99 +0,0 @@
-# Skill Runtime Profile
-
-Status: 1.0 release-candidate profile
-
-The `skill_runtime` profile models a governed skill, capability module or
-service as a Mirai Graph-managed runtime context.
-
-It is designed for systems where the graph helps with routing, capability
-selection, policy reminders, readiness, evidence and generated context, while
-raw source material remains authoritative for detailed domain meaning.
-
-## Object Kinds
-
-The profile includes:
-
-- `skill`;
-- `capability`;
-- `activity`;
-- `rule`;
-- `owner_domain`;
-- `source_boundary`;
-- `raw_source`;
-- `runtime_context`;
-- `projection_view`;
-- `quality_gate`;
-- `adoption_report`;
-- `effectiveness_report`;
-- `semantic_review`;
-- `runtime_policy`;
-- `fallback`;
-- `handoff`;
-- `learning_proposal`;
-- `federation_contract`;
-- `governance_gate`;
-- `evidence`.
-
-## Relation Types
-
-The profile includes relation types for:
-
-- owner boundaries;
-- capabilities and activities;
-- rules and quality gates;
-- raw-source fallback;
-- generated runtime context;
-- semantic preservation;
-- effectiveness measurement;
-- routing and handoff;
-- federation contracts;
-- learning proposals.
-
-## Hybrid SOT Boundary
-
-This profile uses [Hybrid Source Of Truth](../../standard/hybrid-source-of-truth.md).
-
-```text
-graph context can route and orient work
-raw source remains authoritative for domain meaning
-```
-
-Generated runtime context must not rewrite skill sources or grant canonical
-write permission.
-
-## Runtime Boundary
-
-Runtime use follows [Runtime Enablement](../../standard/runtime-enablement.md).
-
-Graph-mode can become the primary context only for a declared task and scope
-after runtime preflight. Sensitive work requires fallback to raw sources and
-the relevant approval gates.
-
-The profile does not provide a separate runtime engine. Skills use the public
-[Project Technology](../../standard/project-technology.md) extension for
-inventory, task-scoped context, synchronization and verification. The profile
-continues to own the vocabulary of skill capabilities, rules and relations.
-
-## Federation Routing Control Loop
-
-When a skill runtime participates in a larger federation, routing should be
-auditable through:
-
-- [Route Explanation](../../standard/route-explanation.md);
-- [Routing Fixtures](../../standard/routing-fixtures.md);
-- [Federation Health](../../standard/federation-health.md);
-- [Route Regression](../../standard/route-regression.md).
-
-These artifacts explain and test routing behavior. They do not authorize
-execution or source rewrites.
-
-## Minimal Fixture
-
-The executable fixture is:
-
-```text
-examples/skill-runtime-minimal/
-```
-
-It proves only profile shape conformance. It does not prove semantic
-completeness, operational effectiveness or runtime safety.

package/profiles/skill-runtime/profile.json

@@ -1,58 +0,0 @@
-{
-  "id": "profile.skill_runtime",
-  "name": "Skill Runtime",
-  "version": "0.1.0",
-  "description": "Profile for modeling a governed skill, capability module or service as a Hybrid Source Of Truth runtime context.",
-  "allowed_object_kinds": [
-    "skill",
-    "capability",
-    "activity",
-    "rule",
-    "owner_domain",
-    "source_boundary",
-    "raw_source",
-    "runtime_context",
-    "projection_view",
-    "quality_gate",
-    "adoption_report",
-    "effectiveness_report",
-    "semantic_review",
-    "runtime_policy",
-    "fallback",
-    "handoff",
-    "learning_proposal",
-    "federation_contract",
-    "governance_gate",
-    "evidence"
-  ],
-  "allowed_relation_types": [
-    "owns_domain",
-    "has_capability",
-    "supports_activity",
-    "uses_rule",
-    "requires_gate",
-    "requires_fallback",
-    "reads_source",
-    "generates_context",
-    "projects_view",
-    "evidences",
-    "preserves_semantics",
-    "measures_effectiveness",
-    "routes_to",
-    "hands_off_to",
-    "exports_contract",
-    "federates_with",
-    "proposes_learning",
-    "governs",
-    "related_to"
-  ],
-  "default_readiness": "draft",
-  "governance_gates": [
-    "human_review_before_canonical_write",
-    "semantic_preservation_before_runtime",
-    "effectiveness_before_federation",
-    "runtime_enablement_before_primary_context",
-    "fallback_required_for_sensitive_tasks",
-    "public_safety_before_release"
-  ]
-}

package/profiles/societal-governance/README.md

@@ -1,129 +0,0 @@
-# Societal Governance Profile
-
-Status: proposal/experimental profile
-
-The societal-governance profile models governance systems where humans,
-bounded AI avatars, norms, initiatives, decisions, execution and outcome
-learning form a controlled evolutionary graph.
-
-It extends the existing `organization_governance`, `character_layer`,
-`process_control` and evidence-control patterns. It does not claim legal,
-democratic, cryptographic or public-governance validity.
-
-## Scope
-
-Use this profile for synthetic or bounded shadow-governance models such as:
-
-- organization governance experiments;
-- SIMAI OS shadow governance;
-- non-binding municipal or public-policy simulations;
-- research fixtures for rights, consent, privacy, appeal and learning gates.
-
-Do not use this profile as a production public-governance authority model.
-
-## Typical Objects
-
-- `governance_dna`
-- `governance_level`
-- `participant`
-- `stakeholder_group`
-- `ai_avatar`
-- `avatar_contract`
-- `delegation_scope`
-- `consent_receipt`
-- `private_decision_journal`
-- `initiative`
-- `simulation`
-- `legitimacy_gate`
-- `rights_gate`
-- `privacy_gate`
-- `affectedness_gate`
-- `anti_capture_gate`
-- `participation_signal`
-- `decision`
-- `decision_trace`
-- `execution_mandate`
-- `outcome_evidence`
-- `appeal`
-- `harm_report`
-- `learning_proposal`
-- `amendment_proposal`
-- `evidence`
-
-## Typical Relations
-
-- `defined_by_dna`
-- `member_of_level`
-- `represented_by_avatar`
-- `bound_by_contract`
-- `delegates_scope_to`
-- `has_consent_receipt`
-- `requires_reconsent`
-- `opens_initiative`
-- `affects_group`
-- `requires_gate`
-- `passes_gate`
-- `fails_gate`
-- `uses_simulation`
-- `uses_evidence`
-- `casts_signal`
-- `summarizes_private_proof`
-- `supports_decision_trace`
-- `authorizes_decision`
-- `creates_execution_mandate`
-- `assigned_to_executor`
-- `produces_outcome`
-- `has_appeal_path`
-- `triggers_appeal`
-- `reports_harm`
-- `creates_learning_proposal`
-- `proposes_dna_amendment`
-- `approved_update_to`
-
-## Governance Gates
-
-- Rights before optimization.
-- Consent before avatar action.
-- Re-consent after model, scope or authority change.
-- Privacy before public trace.
-- Affectedness before weighted decision.
-- Anti-capture before weight change.
-- Human owner before binding action.
-- Evidence before execution mandate.
-- Appeal path before governed decision.
-- Outcome evidence before learning update.
-- Generated context is not authorization.
-- Feedback is not automatic graph update.
-
-## Conformance Levels
-
-- `G0_concept`: model is described, no executable trace.
-- `G1_structured_shadow`: non-binding traces on non-sensitive cases.
-- `G2_assisted_internal`: human-owned internal decisions with AI-assisted
-  traces.
-- `G3_governed_internal`: bounded organizational governance with policy, audit,
-  rollback and evidence.
-- `G4_public_simulation`: public-policy simulations with no binding authority.
-- `G5_limited_public_pilot`: legally and ethically reviewed civic pilot.
-- `G6_constitutional_governance`: formal public governance; out of scope for
-  current evidence.
-
-This proposal profile is intended for `G0_concept` and `G1_structured_shadow`
-fixtures until independent review and pilot evidence exist.
-
-## Boundaries
-
-- AI avatars can explain, recommend, summarize and submit signals only inside
-  explicit consent and delegation scopes.
-- AI avatars and generated context cannot authorize decisions.
-- Contribution, expertise or implementation weights cannot override protected
-  rights.
-- Private decision journals must not be exposed in public traces.
-- Decisions require appeal or reopen paths.
-- Outcome evidence creates learning proposals; it does not update canonical
-  graph state automatically.
-- Public examples must remain synthetic and non-binding.
-
-## Example
-
-See [Societal Governance Minimal](../../examples/societal-governance-minimal/README.md).

package/profiles/societal-governance/profile.json

@@ -1,121 +0,0 @@
-{
-  "id": "profile.societal_governance",
-  "name": "Societal Governance",
-  "version": "0.1.0-proposal",
-  "readiness": "proposal_only",
-  "description": "Proposal profile for modeling rights-bounded governance systems with participants, AI avatars, consent, legitimacy gates, decision traces, execution mandates, outcome evidence and controlled learning proposals.",
-  "allowed_object_kinds": [
-    "governance_dna",
-    "governance_level",
-    "participant",
-    "stakeholder_group",
-    "ai_avatar",
-    "avatar_contract",
-    "delegation_scope",
-    "consent_receipt",
-    "private_decision_journal",
-    "initiative",
-    "simulation",
-    "legitimacy_gate",
-    "rights_gate",
-    "privacy_gate",
-    "affectedness_gate",
-    "anti_capture_gate",
-    "participation_signal",
-    "participation_weight",
-    "generated_context",
-    "decision",
-    "decision_trace",
-    "execution_mandate",
-    "execution_owner",
-    "resource_commitment",
-    "rollback_path",
-    "outcome_metric",
-    "outcome_evidence",
-    "appeal",
-    "harm_report",
-    "learning_proposal",
-    "amendment_proposal",
-    "evidence"
-  ],
-  "allowed_relation_types": [
-    "defined_by_dna",
-    "member_of_level",
-    "represented_by_avatar",
-    "bound_by_contract",
-    "delegates_scope_to",
-    "has_consent_receipt",
-    "requires_reconsent",
-    "opens_initiative",
-    "affects_group",
-    "requires_gate",
-    "passes_gate",
-    "fails_gate",
-    "imposes_condition",
-    "uses_simulation",
-    "uses_evidence",
-    "casts_signal",
-    "summarizes_private_proof",
-    "supports_decision_trace",
-    "authorizes_decision",
-    "creates_execution_mandate",
-    "assigned_to_executor",
-    "commits_resource",
-    "has_rollback_path",
-    "produces_outcome",
-    "has_appeal_path",
-    "triggers_appeal",
-    "reports_harm",
-    "creates_learning_proposal",
-    "proposes_dna_amendment",
-    "approved_update_to",
-    "exposes_private_trace",
-    "overrides_rights",
-    "auto_applies_learning"
-  ],
-  "default_readiness": "draft",
-  "governance_gates": [
-    "rights_before_optimization",
-    "consent_before_avatar_action",
-    "reconsent_after_model_or_scope_change",
-    "privacy_before_public_trace",
-    "affectedness_before_weighted_decision",
-    "anti_capture_before_weight_change",
-    "human_owner_before_binding_action",
-    "evidence_before_execution_mandate",
-    "appeal_path_before_governed_decision",
-    "outcome_evidence_before_learning_update",
-    "no_ai_avatar_as_decision_authority",
-    "no_generated_context_as_authorization",
-    "no_feedback_as_automatic_update"
-  ],
-  "extends_profiles": [
-    "organization_governance",
-    "character_layer",
-    "process_control",
-    "implementation_control"
-  ],
-  "public_safety": {
-    "allowed": [
-      "synthetic examples",
-      "shadow-mode organizational pilots",
-      "non-binding simulations"
-    ],
-    "disallowed": [
-      "real citizen records",
-      "private decision journals in public traces",
-      "binding public-governance claims",
-      "legal validity claims",
-      "automatic learning updates"
-    ]
-  },
-  "conformance_levels": [
-    "G0_concept",
-    "G1_structured_shadow",
-    "G2_assisted_internal",
-    "G3_governed_internal",
-    "G4_public_simulation",
-    "G5_limited_public_pilot",
-    "G6_constitutional_governance"
-  ]
-}

package/profiles/software-specification/README.md

@@ -1,142 +0,0 @@
-# Software Specification Profile
-
-Status: 1.0 release-candidate profile
-
-## Purpose
-
-The software-specification profile adapts Mirai Graph to software packages,
-features, requirements, constraints, risks, implementation tasks and evidence.
-
-It helps connect:
-
-- package scope;
-- requirements;
-- features;
-- components;
-- constraints;
-- risks;
-- implementation tasks;
-- decisions;
-- evidence.
-
-## Object Kinds
-
-Initial allowed object kinds:
-
-- `package`;
-- `feature`;
-- `requirement`;
-- `constraint`;
-- `risk`;
-- `component`;
-- `task`;
-- `decision`;
-- `evidence`.
-
-## Relation Types
-
-Initial allowed relation types:
-
-- `implements`;
-- `depends_on`;
-- `blocks`;
-- `evidences`;
-- `generates`;
-- `related_to`;
-- `supersedes`.
-
-## Governance Gates
-
-Initial gates:
-
-- `human_review_before_canonical_write`;
-- `public_safety_before_release`.
-
-Domain-specific projects may add gates such as:
-
-- `requirement_traceability_before_implementation`;
-- `security_review_before_release`;
-- `acceptance_criteria_before_development`;
-- `test_evidence_before_acceptance`.
-
-## Traceability Guidance
-
-For implementation work, prefer explicit task traceability:
-
-```text
-task -> implements -> feature
-task -> depends_on -> requirement
-feature -> implements -> requirement
-component -> implements -> feature
-constraint -> blocks -> risk
-decision -> blocks -> risk
-```
-
-This makes generated context packs less likely to omit the underlying
-requirement when the task title only matches an implementation detail.
-
-## Readiness Guidance
-
-Use:
-
-- `draft` for early requirements, components, decisions and task plans;
-- `review_ready` when a requirement, feature, task or risk is ready for owner
-  review;
-- `accepted` only after review evidence exists;
-- `blocked` when missing requirement, security, acceptance or test evidence
-  prevents progress;
-- `deprecated` when a requirement, feature or decision is superseded.
-
-## Recommended Minimal Package
-
-A useful Level 1 software-specification package should include:
-
-- at least one `package`;
-- at least one `requirement`;
-- at least one `feature`;
-- at least one implementation `task`;
-- at least one `evidence` object;
-- direct traceability from task to feature and requirement;
-- constraints and risks when safety, privacy, security or reliability matters;
-- gate results for public safety or requirement traceability when the package
-  is used as evidence.
-
-## Current Limitations
-
-The profile does not yet define first-class object kinds for:
-
-- API endpoints;
-- acceptance criteria;
-- test cases;
-- verification results;
-- non-functional requirement categories;
-- owners or reviewers.
-
-Use `requirement`, `constraint`, `task`, `decision` and `evidence` until these
-extensions are standardized.
-
-## Adoption Path
-
-Recommended path:
-
-1. Start with a graph seed.
-2. Draft package, requirement, feature, component, task, risk and evidence
-   objects.
-3. Add direct task-to-requirement traceability before using the graph for
-   implementation context.
-4. Add constraints and risks for security, reliability or privacy concerns.
-5. Run package validation and readiness scoring.
-6. Generate a context pack for one implementation task.
-7. Review omissions before using the context pack for work.
-
-## Pilot Evidence
-
-Pilot 002 uses this profile:
-
-`pilots/independent-implementation-002-software-specification/`
-
-Pilot finding:
-
-The profile is sufficient for a small Level 1 package, but needs richer
-modeling for acceptance criteria, API contracts, non-functional requirements,
-test evidence and conformance fixtures.

package/profiles/software-specification/profile.json

@@ -1,31 +0,0 @@
-{
-  "id": "profile.software_specification",
-  "name": "Software Specification",
-  "version": "0.1.0",
-  "description": "Profile for modeling software packages, features, constraints, risks and evidence.",
-  "allowed_object_kinds": [
-    "package",
-    "feature",
-    "requirement",
-    "constraint",
-    "risk",
-    "component",
-    "task",
-    "decision",
-    "evidence"
-  ],
-  "allowed_relation_types": [
-    "implements",
-    "depends_on",
-    "blocks",
-    "evidences",
-    "generates",
-    "related_to",
-    "supersedes"
-  ],
-  "default_readiness": "draft",
-  "governance_gates": [
-    "human_review_before_canonical_write",
-    "public_safety_before_release"
-  ]
-}

package/publications/README.md

@@ -1,39 +0,0 @@
-# Publications
-
-This directory tracks public publications, preprints, citation metadata and
-reproducibility notes related to Mirai Graph.
-
-## Current Status
-
-The model is being prepared as an open research and standardization project.
-Publication links should be added only when the public page, DOI or venue record
-is confirmed.
-
-Do not describe a preprint as peer reviewed unless the venue record explicitly
-states peer-reviewed publication status.
-
-For the current public evidence framing, see
-`docs/research/scientific-evidence-package.md`.
-
-## Planned Records
-
-- initial preprint record;
-- preprint outline;
-- [Semantic Intent And Goal Vector Control](semantic-intent-goal-vector-control/README.md);
-- [Observable Self-Correcting AI Workflows](observable-self-correcting-ai-workflows/README.md);
-- public citation note;
-- reproducibility package note;
-- future journal manuscript records.
-
-## Article Candidates
-
-The current `1.0.0-rc.6` release candidate supports two public-safe article
-tracks:
-
-| Candidate | Focus | Current evidence level |
-| --- | --- | --- |
-| Semantic Intent And Goal Vector Control | preventing wrong-start and false-completion drift before and after work execution | public standard, schema, synthetic examples and negative fixtures |
-| Observable Self-Correcting AI Workflows | explaining what happened, checking process conformance and routing corrections into Kaizen/replay | public standard sections, schemas, synthetic examples and negative fixtures |
-
-Both tracks still require sanitized pilot/replay evidence before broad
-effectiveness claims.