mirai-graph 1.3.0 → 2.1.0

package/docs/transfer/ai-employee-profile-proposal.md

@@ -1,119 +0,0 @@
-# AI Employee Profile Proposal
-
-Status: alpha proposal
-
-## Purpose
-
-This proposal defines the public Mirai Graph direction for modeling an
-AI employee as a governed runtime over an evolutionary graph.
-
-The goal is to support systems such as Larena/AI without turning Mirai Graph
-into a single AI-agent framework or publishing private implementation details.
-
-## Core Thesis
-
-```text
-AI employee is not a prompt.
-AI employee is a governed runtime over an evolutionary graph.
-The graph is the canonical brain/state/control model.
-The runtime executes through tools, policies, approvals and audit.
-```
-
-## Why A Separate Profile
-
-The existing profiles model projects and software specifications. An
-AI-employee system needs additional first-class concepts:
-
-- roles and departments;
-- skills and specializations;
-- knowledge and memory;
-- tools and capabilities;
-- workflows and automation rules;
-- policies and approval gates;
-- actions, results, feedback and lessons;
-- federation endpoints for multi-employee or multi-service systems.
-
-These concepts should not bloat the core standard. They belong in a profile
-that specializes Mirai Graph for agentic work while preserving the core
-governance model.
-
-## Public Scope
-
-The public standard may define:
-
-- graph object kinds for AI-employee modeling;
-- relation types for skill, tool, policy and delegation structure;
-- canonical state versus runtime/event boundaries;
-- feedback learning gates;
-- synthetic fixtures;
-- validation rules;
-- adoption guidance.
-
-The public standard must not define:
-
-- private Larena package internals;
-- SIMAI customer or employee data;
-- secret handling details beyond public-safe policy boundaries;
-- one mandatory LLM provider;
-- one mandatory runtime or queue implementation;
-- uncontrolled self-modifying behavior.
-
-## Larena/AI Boundary
-
-Mirai Graph should define the portable model. Larena/AI may implement runtime
-execution, UI, storage, queues, tools, permissions, audit logs and package
-integration.
-
-Portable Mirai Graph layer:
-
-- AI employee graph model;
-- object and relation semantics;
-- policy gates;
-- feedback learning lifecycle;
-- profile conformance fixtures.
-
-Larena/AI runtime layer:
-
-- authentication and authorization;
-- tool adapters;
-- execution queues;
-- event logs;
-- UI screens;
-- deployment topology;
-- provider-specific integration.
-
-## Minimal Alpha Scope
-
-The first public alpha should include:
-
-- `profiles/ai-employee/profile.json`;
-- `profiles/ai-employee/README.md`;
-- `standard/graph-as-ai-brain.md`;
-- `standard/action-runtime-boundary.md`;
-- `standard/feedback-learning-gate.md`;
-- `docs/adoption/ai-employee-graph.md`;
-- `examples/ai-employee-minimal/`;
-- validation through `npm run release:check`.
-
-## Safety Principles
-
-1. Feedback does not mutate canonical graph state directly.
-2. Generated context does not become authorization.
-3. Runtime decisions do not equal executed actions.
-4. Executed actions need audited results.
-5. High-risk actions require explicit policy and approval gates.
-6. Learning requires proposal, replay or evaluation, approval and canonical
-   update.
-7. Federation endpoints need compatibility and trust boundaries.
-
-## Claim Rule
-
-This proposal supports only this claim:
-
-```text
-Mirai Graph can model the public structure of an AI employee as a governed graph
-profile with explicit runtime, action and learning boundaries.
-```
-
-It does not prove that any runtime implementation is safe, autonomous or
-effective.

package/docs/transfer/federation-routing-control-loop-2026-05-31.md

@@ -1,28 +0,0 @@
-# Federation Routing Control Loop Transfer
-
-Date: 2026-05-31
-
-Status: public-safe transfer note
-
-## Transferred Concepts
-
-This transfer generalizes internal federation routing work into public-safe
-Mirai Graph standard drafts:
-
-- route explanation;
-- routing fixtures;
-- forbidden route shapes;
-- federation health dashboard;
-- route regression result;
-- high-risk route preflight boundary.
-
-## Public-Safe Boundary
-
-This transfer does not publish private skill playbooks, customer routes,
-runtime paths, credentials, internal metrics or owner-specific methods.
-
-## Claim Boundary
-
-The transferred artifacts support schema and standard shape claims only. They
-do not prove real-world productivity, runtime safety, scientific validity or
-graph-only source-of-truth readiness.

package/docs/transfer/graph-capability-map-2026-05-29.md

@@ -1,126 +0,0 @@
-# Mirai Graph Public Capability Map
-
-Date: 2026-05-29
-
-Status: public-safe transfer map
-
-## Purpose
-
-This map connects practical `$graph` capability areas to public Mirai Graph
-standard sections, schemas, tooling, examples and evidence artifacts.
-
-It is not a publication of internal skill implementation details. It records
-which reusable ideas are already represented in the open repository and which
-ideas still need public standardization.
-
-## Public Transfer Boundary
-
-Transfer only public-safe concepts:
-
-- implementation-independent model elements;
-- schemas and validators;
-- synthetic examples and benchmarks;
-- adoption guides;
-- research protocols;
-- limited claims backed by public artifacts.
-
-Do not transfer:
-
-- private project data;
-- raw internal chats or handoffs;
-- customer or employee material;
-- secrets or private runtime paths;
-- organization-specific procedures that have not been generalized.
-
-## Capability Coverage
-
-| Practical capability area | Public Mirai Graph coverage | Status | Next public step |
-| --- | --- | --- | --- |
-| Repository as graph source of truth | `docs/repository-purpose.md`, `README.md`, `ROADMAP.md` | covered | Keep purpose, roadmap and release notes aligned. |
-| Core object model | `standard/object-model.md`, `schemas/object.schema.json`, package fixtures | covered with notes | Add richer object-family examples for more profiles. |
-| Typed relation model | `standard/relation-model.md`, `schemas/relation.schema.json`, relation-id validation | covered with notes | Expand impact traversal and change-review semantics. |
-| Evidence and provenance | `standard/evidence-provenance.md`, evidence fields, synthetic source corpus | covered with notes | Add evidence-level taxonomy and validation checks. |
-| Lifecycle and readiness | `standard/lifecycle-readiness.md`, readiness validation, `packages/cli/readiness-score.js` | partial | Standardize readiness-score bands and caps as a first-class section. |
-| Graph DNA | `standard/graph-dna.md` | partial | Add validation examples and DNA-to-seed traceability. |
-| Graph Seed | `standard/graph-seed.md`, `schemas/graph-seed.schema.json`, seed examples, seed validation | covered with notes | Add more negative seed fixtures and adoption walkthroughs. |
-| Graph Embryo | `standard/graph-embryo.md`, `schemas/graph-embryo.schema.json`, seed preview output | covered with notes | Add embryo review-result artifacts and promotion rules. |
-| Canonical graph state | `standard/standard-v0.1.md`, package manifest schema, graph objects/relations | covered with notes | Define canonical write/update proposal format. |
-| Generated context packs | `standard/context-pack-generation.md`, `schemas/context-pack.schema.json`, `packages/cli/context-pack.js`, context-pack validation | covered with notes | Add schema-level fixtures for selection explanation edge cases. |
-| Selection explanations | context-pack `selection.object_explanations` and `selection.relation_explanations` | covered | Add reviewer guidance for interpreting scores. |
-| Safety-sensitive context inclusion | context-pack generation policy and semantic review artifacts | partial | Turn safety inclusion into an explicit standard subsection. |
-| Governance gates | `standard/governance.md`, gate result schema and examples | partial | Add proposal/approval flow with dry-run and apply boundaries. |
-| Profiles | `standard/profiles.md`, `profiles/`, profile validation and profile conformance fixtures | covered with notes | Add organization-governance, research-program and human-AI profiles later. |
-| AI employee graph-as-brain | `profiles/ai-employee/`, `standard/graph-as-ai-brain.md`, `examples/ai-employee-minimal/` | alpha | Validate through Larena/AI pilot before making stronger runtime claims. |
-| Action/runtime boundary | `standard/action-runtime-boundary.md`, AI employee fixture approval gate | alpha | Add proposal/action/result schemas after runtime patterns stabilize. |
-| Feedback learning gate | `standard/feedback-learning-gate.md`, AI employee fixture feedback and lesson proposal | alpha | Add replay/evaluation result artifact schema. |
-| Profile conformance evidence | positive and negative fixtures, `validate-profile-results.js` | covered | Keep stored results regenerated before releases. |
-| Test suite and conformance levels | `standard/test-suite.md`, `standard/conformance-levels.md`, `npm run release:check` | covered with notes | Split release checks into named conformance levels. |
-| Context reduction benchmark | `benchmarks/synthetic-context-reduction-v0/` and result artifacts | covered with limits | Add more benchmark tasks and ablation protocols. |
-| Semantic completeness review | `docs/research/semantic-completeness-review-protocol.md`, review JSON artifacts | partial | Add independent reviewer workflow and disagreement handling. |
-| Measure-convert-optimize cycle | `benchmarks/protocols/measure-convert-optimize.md` | partial | Add a second public benchmark and repeatable result writer. |
-| Adoption path | `docs/adoption/getting-started.md`, `docs/adoption/seed-to-validated-package.md` | covered with notes | Add one complete tutorial from seed to context-pack review. |
-| Federation / cross-graph contracts | not yet public | planned | Draft `standard/federation.md` after core package contracts stabilize. |
-| Runtime enablement policy | represented only indirectly through governance and readiness | planned | Draft public-safe runtime enablement policy without internal runtime details. |
-| Migration/versioning | not yet public as a standard section | planned | Draft `standard/versioning-migrations.md` and migration dry-run artifacts. |
-
-## Current Public Maturity
-
-Current repository maturity:
-
-`public_alpha_standard_with_executable_fixtures`
-
-Reason:
-
-The repository now contains a public standard foundation, schemas, validators,
-context-pack generation, readiness scoring, profiles, synthetic benchmark,
-semantic review artifacts, positive and negative conformance fixtures, and
-release checks.
-
-The repository is not yet:
-
-- a complete standard;
-- a full reference implementation;
-- proof of real-world productivity improvement;
-- a runtime action framework;
-- a publication-ready empirical evidence package.
-
-## Priority Standardization Backlog
-
-### P0: Prepare `v0.1.0-alpha.4`
-
-- summarize completed executable standard improvements;
-- publish release notes;
-- keep claims limited to public alpha tooling and synthetic evidence.
-
-### P1: Human Control And Governance
-
-- proposal/approval flow;
-- dry-run/apply boundary;
-- governance gate result lifecycle;
-- safety and public-release gates.
-
-### P2: Readiness And Review
-
-- readiness score standard section;
-- score bands and caps;
-- semantic review workflow;
-- reviewer disagreement and limitation records.
-
-### P3: Versioning And Federation
-
-- schema versioning;
-- migration reports;
-- cross-graph compatibility;
-- public federation contract.
-
-## Claim Rule
-
-This capability map supports only this claim:
-
-```text
-Mirai Graph has a public alpha mapping from practical graph-method capabilities
-to open standard sections, schemas, tooling and public-safe evidence artifacts.
-```
-
-It does not claim that the practical `$graph` skill is fully public, fully
-standardized or externally validated.

package/docs/transfer/graph-skill-to-mirai-graph.md

@@ -1,128 +0,0 @@
-# From `$graph` Skill To Mirai Graph
-
-Status: initial transfer map
-
-## Purpose
-
-This note defines how reusable knowledge from the practical `$graph` skill
-should move into the public Mirai Graph repository.
-
-## Transfer Rule
-
-`$graph` is the practical methodology owner. Mirai Graph is the public standard
-and adoption layer.
-
-Transfer a concept only when it is:
-
-- reusable outside the original skill;
-- safe for public release;
-- not tied to private SIMAI or customer material;
-- useful for standards, tooling, profiles, examples or research evidence.
-
-## Transfer Classes
-
-### Standard Concept
-
-Belongs in:
-
-- `standard/`
-- `schemas/`
-
-Examples:
-
-- object model;
-- relation model;
-- lifecycle/readiness;
-- evidence/provenance;
-- governance gates;
-- conformance levels.
-
-### Profile Concept
-
-Belongs in:
-
-- `standard/profiles.md`
-- future `profiles/`
-
-Examples:
-
-- software specification profile;
-- project management profile;
-- organization governance profile;
-- research program profile.
-
-### Reference Tooling
-
-Belongs in:
-
-- `packages/`
-- `schemas/`
-- `examples/`
-
-Examples:
-
-- validators;
-- generators;
-- benchmark runners;
-- migration helpers.
-
-### Research Artifact
-
-Belongs in:
-
-- `benchmarks/`
-- `publications/`
-- `docs/research/`
-
-Examples:
-
-- experiment protocol;
-- result note;
-- reproducibility package;
-- validation roadmap.
-
-### Private Or Local Material
-
-Does not belong in public Mirai Graph.
-
-Examples:
-
-- raw internal project materials;
-- internal chats;
-- private skill implementation details not approved for release;
-- customer or employee data;
-- secrets;
-- private metrics without redaction and approval.
-
-## Initial Mapping
-
-| `$graph` Area | Mirai Graph Destination | Notes |
-| --- | --- | --- |
-| Graph Seed | `standard/`, `docs/adoption/` | Should become a public seed method. |
-| Graph Embryo | `standard/`, `docs/adoption/` | Useful for staged graph growth. |
-| Lifecycle/readiness | `standard/lifecycle-readiness.md` | Already started. |
-| Relation/impact model | `standard/relation-model.md` | Needs richer impact semantics. |
-| AI context generation | `standard/context-pack-generation.md` | Needs schema and generator. |
-| Measure-convert-optimize | `docs/research/`, `benchmarks/` | Core validation methodology. |
-| Graph method evolution | `docs/research/` | Core practice-to-science loop. |
-| Federation contracts | future `standard/federation.md` | Later stage. |
-| Runtime policies | future `standard/governance.md` expansion | Public-safe subset only. |
-
-## Next Transfer Batch
-
-1. Draft graph DNA, graph seed and graph embryo public standard sections.
-2. Create graph seed and graph embryo schemas.
-3. Extend the CLI with seed validation.
-4. Draft measure-convert-optimize experiment protocol.
-5. Add profile manifests for the first public profiles.
-
-Current gap report:
-
-`docs/transfer/standard-gap-report-2026-05-28.md`
-
-Current capability map:
-
-`docs/transfer/graph-capability-map-2026-05-29.md`
-
-The dated capability map is the current public-safe transfer state. The
-2026-05-28 gap report is retained as a historical baseline.