mirai-graph 1.4.0 → 2.1.0

package/pilots/independent-implementation-002-software-specification/reports/pilot-report.md

@@ -1,142 +0,0 @@
-# Pilot Report: Independent Implementation 002
-
-Date: 2026-05-28
-
-Status: `pass_with_notes`
-
-## Purpose
-
-Test whether a valid Mirai Graph Level 1 package can be created for a small
-software specification workflow using only the public repository materials.
-
-## Domain
-
-Synthetic team notification API workflow.
-
-No private company, customer, employee, repository issue or production material
-was used.
-
-## Research Question
-
-Can a developer create a valid Level 1 Mirai Graph package for a small software
-specification using only public Mirai Graph docs, schemas and examples?
-
-## Result
-
-Verdict: `pass_with_notes`
-
-The pilot produced:
-
-- valid graph seed;
-- valid Mirai Graph package;
-- valid object and relation files;
-- public-safety gate result;
-- requirement-traceability gate result;
-- generated graph embryo preview;
-- generated context pack;
-- readiness score.
-
-## Evidence
-
-Validation artifacts:
-
-- `results/seed-validation.json`
-- `results/package-validation.json`
-- `results/graph-embryo.json`
-- `results/context-pack.json`
-- `results/readiness-score.json`
-- `reviews/semantic-completeness-review.json`
-
-Package summary:
-
-- objects: `11`;
-- relations: `10`;
-- evidence objects: `1`;
-- accepted objects: `2`;
-- gate results: `2`;
-- target mode: `pilot`;
-- readiness score: `69`;
-- readiness band: `G3_pilot_ready`.
-- semantic completeness verdict: `pass_with_notes`;
-- semantic completeness score: `15/16`.
-
-## What Worked
-
-- The software-specification profile was sufficient for a small package,
-  requirement, feature, component, risk, decision and implementation-task map.
-- The package validator caught structural risks while the graph was drafted.
-- The context-pack generator produced a task-bounded package for
-  `task.implement_retry_policy`.
-- A direct task-to-requirement relation improved traceability and made the
-  generated task context more useful.
-- Semantic review made safety-context omissions explicit.
-- Context-pack selection explanations now show relevance scores and inclusion
-  reasons for selected objects and relations.
-- Readiness scoring correctly capped the result because evidence is synthetic
-  and benchmark evidence is missing for this pilot.
-
-## Gaps Found
-
-### Standard Gap
-
-The software-specification profile is still thin. It validates the basic
-object families but does not yet explain:
-
-- acceptance criteria modeling;
-- API contract or endpoint modeling;
-- non-functional requirements;
-- task-to-requirement traceability expectations;
-- security constraint coverage;
-- test-case and verification-result objects.
-
-### Tooling Gap
-
-The context-pack generator selected a useful task context after explicit
-traceability was added. It still does not:
-
-- explain why each object or relation was selected;
-- score relevance;
-- detect that a task lacks direct requirement traceability;
-- identify omitted safety constraints that may still be important.
-
-### Research Gap
-
-The pilot proves package creation and validation for a synthetic software
-specification. It does not measure:
-
-- implementation speed;
-- defect reduction;
-- reviewer comprehension;
-- downstream AI answer quality;
-- real software delivery outcomes.
-
-### Documentation Gap
-
-The software-specification profile needs a public profile guide with worked
-examples for requirements, features, components, tasks, risks, constraints,
-acceptance criteria and verification artifacts.
-
-## Interpretation
-
-This pilot supports a limited claim:
-
-```text
-A valid Mirai Graph Level 1 package can be created from public repository
-materials for a small synthetic software specification.
-```
-
-It does not support claims about:
-
-- real-world software productivity;
-- semantic completeness;
-- external validity;
-- runtime safety;
-- universal applicability.
-
-## Recommended Next Actions
-
-1. Add software-specification profile documentation.
-2. Add profile conformance fixtures for required traceability patterns.
-3. Add context-pack selection explanations and relevance scores.
-4. Add acceptance-criteria and verification-result modeling examples.
-5. Add a policy for when safety constraints must be included in task context.

package/pilots/independent-implementation-002-software-specification/results/context-pack.json

@@ -1,184 +0,0 @@
-{
-  "id": "context_pack.task_implement_retry_policy",
-  "task_id": "task.implement_retry_policy",
-  "source_graph": "package.notification_api_spec_pilot_002",
-  "generated_at": "2026-05-28T19:46:09.425Z",
-  "included_objects": [
-    "source.notification_api_brief",
-    "requirement.notify_after_event",
-    "feature.webhook_delivery",
-    "feature.retry_policy",
-    "component.delivery_worker",
-    "constraint.no_secret_logging",
-    "risk.duplicate_delivery",
-    "risk.credential_leak",
-    "decision.idempotency_key_required",
-    "task.implement_retry_policy"
-  ],
-  "included_relations": [
-    "relation.feature.webhook_delivery.implements.requirement.notify_after_event",
-    "relation.feature.retry_policy.depends_on.feature.webhook_delivery",
-    "relation.component.delivery_worker.implements.feature.retry_policy",
-    "relation.constraint.no_secret_logging.blocks.risk.credential_leak",
-    "relation.feature.retry_policy.related_to.risk.duplicate_delivery",
-    "relation.decision.idempotency_key_required.blocks.risk.duplicate_delivery",
-    "relation.task.implement_retry_policy.implements.feature.retry_policy",
-    "relation.task.implement_retry_policy.depends_on.requirement.notify_after_event",
-    "relation.task.implement_retry_policy.depends_on.decision.idempotency_key_required"
-  ],
-  "evidence": [
-    "source.notification_api_brief"
-  ],
-  "selection": {
-    "method": "token_matching_one_hop_relation_expansion_safety_shared_evidence",
-    "task_tokens": [
-      "task",
-      "implement",
-      "retry",
-      "policy"
-    ],
-    "object_explanations": [
-      {
-        "id": "source.notification_api_brief",
-        "relevance_score": -1,
-        "reasons": [
-          "evidence referenced by selected object"
-        ]
-      },
-      {
-        "id": "requirement.notify_after_event",
-        "relevance_score": 0,
-        "reasons": [
-          "one-hop relation expansion from relation.task.implement_retry_policy.depends_on.requirement.notify_after_event"
-        ]
-      },
-      {
-        "id": "feature.webhook_delivery",
-        "relevance_score": 0,
-        "reasons": [
-          "one-hop relation expansion from relation.feature.retry_policy.depends_on.feature.webhook_delivery"
-        ]
-      },
-      {
-        "id": "feature.retry_policy",
-        "relevance_score": 6,
-        "reasons": [
-          "exact token match: retry",
-          "exact token match: policy"
-        ]
-      },
-      {
-        "id": "component.delivery_worker",
-        "relevance_score": 0,
-        "reasons": [
-          "one-hop relation expansion from relation.component.delivery_worker.implements.feature.retry_policy"
-        ]
-      },
-      {
-        "id": "constraint.no_secret_logging",
-        "relevance_score": 0,
-        "reasons": [
-          "safety-sensitive object with shared evidence"
-        ]
-      },
-      {
-        "id": "risk.duplicate_delivery",
-        "relevance_score": 0,
-        "reasons": [
-          "one-hop relation expansion from relation.feature.retry_policy.related_to.risk.duplicate_delivery"
-        ]
-      },
-      {
-        "id": "risk.credential_leak",
-        "relevance_score": 0,
-        "reasons": [
-          "safety-sensitive object with shared evidence"
-        ]
-      },
-      {
-        "id": "decision.idempotency_key_required",
-        "relevance_score": 0,
-        "reasons": [
-          "one-hop relation expansion from relation.task.implement_retry_policy.depends_on.decision.idempotency_key_required"
-        ]
-      },
-      {
-        "id": "task.implement_retry_policy",
-        "relevance_score": 12,
-        "reasons": [
-          "exact token match: task",
-          "exact token match: implement",
-          "exact token match: retry",
-          "exact token match: policy"
-        ]
-      }
-    ],
-    "relation_explanations": [
-      {
-        "id": "relation.feature.webhook_delivery.implements.requirement.notify_after_event",
-        "reason": "connects selected objects after safety/evidence expansion",
-        "source": "feature.webhook_delivery",
-        "target": "requirement.notify_after_event"
-      },
-      {
-        "id": "relation.feature.retry_policy.depends_on.feature.webhook_delivery",
-        "reason": "one-hop relation expansion from direct object match",
-        "source": "feature.retry_policy",
-        "target": "feature.webhook_delivery"
-      },
-      {
-        "id": "relation.component.delivery_worker.implements.feature.retry_policy",
-        "reason": "one-hop relation expansion from direct object match",
-        "source": "component.delivery_worker",
-        "target": "feature.retry_policy"
-      },
-      {
-        "id": "relation.constraint.no_secret_logging.blocks.risk.credential_leak",
-        "reason": "connects selected objects after safety/evidence expansion",
-        "source": "constraint.no_secret_logging",
-        "target": "risk.credential_leak"
-      },
-      {
-        "id": "relation.feature.retry_policy.related_to.risk.duplicate_delivery",
-        "reason": "one-hop relation expansion from direct object match",
-        "source": "feature.retry_policy",
-        "target": "risk.duplicate_delivery"
-      },
-      {
-        "id": "relation.decision.idempotency_key_required.blocks.risk.duplicate_delivery",
-        "reason": "connects selected objects after safety/evidence expansion",
-        "source": "decision.idempotency_key_required",
-        "target": "risk.duplicate_delivery"
-      },
-      {
-        "id": "relation.task.implement_retry_policy.implements.feature.retry_policy",
-        "reason": "one-hop relation expansion from direct object match",
-        "source": "task.implement_retry_policy",
-        "target": "feature.retry_policy"
-      },
-      {
-        "id": "relation.task.implement_retry_policy.depends_on.requirement.notify_after_event",
-        "reason": "one-hop relation expansion from direct object match",
-        "source": "task.implement_retry_policy",
-        "target": "requirement.notify_after_event"
-      },
-      {
-        "id": "relation.task.implement_retry_policy.depends_on.decision.idempotency_key_required",
-        "reason": "one-hop relation expansion from direct object match",
-        "source": "task.implement_retry_policy",
-        "target": "decision.idempotency_key_required"
-      }
-    ]
-  },
-  "assumptions": [
-    "Alpha generator uses token matching plus one-hop relation expansion.",
-    "Safety-sensitive constraints and risks with shared evidence are included."
-  ],
-  "omissions": [
-    "package.team_notification_api"
-  ],
-  "limitations": [
-    "Generated context pack is not canonical state.",
-    "This alpha generator does not prove semantic completeness."
-  ]
-}

package/pilots/independent-implementation-002-software-specification/results/graph-embryo.json

@@ -1,90 +0,0 @@
-{
-  "schema_version": "0.1.0",
-  "embryo_id": "embryo.notification_api_spec",
-  "seed_id": "seed.notification_api_spec",
-  "generated_at": "2026-05-28T18:22:20.166Z",
-  "candidate_objects": [
-    {
-      "id": "package.candidate_1",
-      "kind": "package",
-      "title": "Candidate Package",
-      "summary": "Candidate package object proposed from seed seed.notification_api_spec.",
-      "readiness": "draft",
-      "evidence": [],
-      "profile": "software_specification"
-    },
-    {
-      "id": "feature.candidate_2",
-      "kind": "feature",
-      "title": "Candidate Feature",
-      "summary": "Candidate feature object proposed from seed seed.notification_api_spec.",
-      "readiness": "draft",
-      "evidence": [],
-      "profile": "software_specification"
-    },
-    {
-      "id": "requirement.candidate_3",
-      "kind": "requirement",
-      "title": "Candidate Requirement",
-      "summary": "Candidate requirement object proposed from seed seed.notification_api_spec.",
-      "readiness": "draft",
-      "evidence": [],
-      "profile": "software_specification"
-    },
-    {
-      "id": "constraint.candidate_4",
-      "kind": "constraint",
-      "title": "Candidate Constraint",
-      "summary": "Candidate constraint object proposed from seed seed.notification_api_spec.",
-      "readiness": "draft",
-      "evidence": [],
-      "profile": "software_specification"
-    },
-    {
-      "id": "risk.candidate_5",
-      "kind": "risk",
-      "title": "Candidate Risk",
-      "summary": "Candidate risk object proposed from seed seed.notification_api_spec.",
-      "readiness": "draft",
-      "evidence": [],
-      "profile": "software_specification"
-    },
-    {
-      "id": "component.candidate_6",
-      "kind": "component",
-      "title": "Candidate Component",
-      "summary": "Candidate component object proposed from seed seed.notification_api_spec.",
-      "readiness": "draft",
-      "evidence": [],
-      "profile": "software_specification"
-    },
-    {
-      "id": "task.candidate_7",
-      "kind": "task",
-      "title": "Candidate Task",
-      "summary": "Candidate task object proposed from seed seed.notification_api_spec.",
-      "readiness": "draft",
-      "evidence": [],
-      "profile": "software_specification"
-    },
-    {
-      "id": "decision.candidate_8",
-      "kind": "decision",
-      "title": "Candidate Decision",
-      "summary": "Candidate decision object proposed from seed seed.notification_api_spec.",
-      "readiness": "draft",
-      "evidence": [],
-      "profile": "software_specification"
-    }
-  ],
-  "candidate_relations": [],
-  "uncertain_items": [
-    "This is a deterministic seed preview, not extracted canonical graph state."
-  ],
-  "review_questions": [
-    "Do the candidate object families match the graph purpose?",
-    "Are source boundaries sufficient for the first extraction pass?",
-    "Should canonical write remain blocked until human review?"
-  ],
-  "stop_conditions_triggered": []
-}

package/pilots/independent-implementation-002-software-specification/results/package-validation.json

@@ -1,7 +0,0 @@
-{
-  "mode": "package",
-  "target": "pilots/independent-implementation-002-software-specification",
-  "valid": true,
-  "errors": [],
-  "warnings": []
-}

package/pilots/independent-implementation-002-software-specification/results/profile-conformance-result.json

@@ -1,28 +0,0 @@
-{
-  "schema_version": "0.1.0",
-  "result_id": "profile_conformance.package.notification_api_spec_pilot_002",
-  "checked_at": "2026-05-29",
-  "fixture": "pilots/independent-implementation-002-software-specification/",
-  "package_id": "package.notification_api_spec_pilot_002",
-  "profile": "software_specification",
-  "conformance_level": "level_1",
-  "validator": "packages/cli/validate-mirai-graph.js",
-  "command": "node packages/cli/validate-mirai-graph.js pilots/independent-implementation-002-software-specification",
-  "valid": true,
-  "errors": [],
-  "warnings": [],
-  "checks": [
-    "required package, object, relation and gate-result fields",
-    "unique object and relation ids",
-    "relation source and target existence",
-    "relation id consistency",
-    "readiness and gate verdict values",
-    "software_specification object-kind conformance",
-    "software_specification relation-type conformance"
-  ],
-  "claim_supported": "The package follows the declared Mirai Graph profile shape checked by the current validator.",
-  "limitations": [
-    "This result does not prove semantic completeness, usefulness or downstream task performance.",
-    "The pilot package is synthetic public-safe evidence."
-  ]
-}

package/pilots/independent-implementation-002-software-specification/results/readiness-score.json

@@ -1,26 +0,0 @@
-{
-  "schema_version": "0.1.0",
-  "target_mode": "pilot",
-  "score": 69,
-  "band": "G3_pilot_ready",
-  "evidence": [
-    "package manifest exists",
-    "objects exist: 11",
-    "relations exist: 10",
-    "evidence objects exist: 1",
-    "accepted objects exist: 2",
-    "gate results exist: 2"
-  ],
-  "blockers": [],
-  "caps": [
-    {
-      "reason": "no benchmark evidence for pilot/runtime target",
-      "max_score": 69
-    },
-    {
-      "reason": "synthetic-only evidence cannot exceed pilot-ready navigation score",
-      "max_score": 69
-    }
-  ],
-  "note": "Release-candidate readiness score; use as navigation, not approval."
-}

package/pilots/independent-implementation-002-software-specification/results/seed-validation.json

@@ -1,7 +0,0 @@
-{
-  "mode": "seed",
-  "target": "pilots/independent-implementation-002-software-specification/graph-seed.json",
-  "valid": true,
-  "errors": [],
-  "warnings": []
-}

package/pilots/independent-implementation-002-software-specification/reviews/semantic-completeness-review.json

@@ -1,81 +0,0 @@
-{
-  "schema_version": "0.1.0",
-  "review_id": "semantic_review.notification_api_spec_pilot_002.task_implement_retry_policy",
-  "reviewed_at": "2026-05-28",
-  "reviewer": "Mirai Graph maintainers",
-  "protocol": "docs/research/semantic-completeness-review-protocol.md",
-  "target": {
-    "package_id": "package.notification_api_spec_pilot_002",
-    "context_pack": "pilots/independent-implementation-002-software-specification/results/context-pack.json",
-    "task_id": "task.implement_retry_policy"
-  },
-  "verdict": "pass_with_notes",
-  "score_total": 15,
-  "score_max": 16,
-  "dimensions": [
-    {
-      "name": "task_relevance",
-      "score": 2,
-      "note": "The context pack is centered on implementing the retry policy."
-    },
-    {
-      "name": "required_object_coverage",
-      "score": 2,
-      "note": "The task, retry feature, webhook feature, delivery worker, requirement, duplicate-delivery risk and idempotency decision are included."
-    },
-    {
-      "name": "required_relation_coverage",
-      "score": 2,
-      "note": "The task keeps direct traceability to the feature, requirement and idempotency decision."
-    },
-    {
-      "name": "evidence_preservation",
-      "score": 2,
-      "note": "The synthetic source brief is preserved."
-    },
-    {
-      "name": "risk_and_blocker_visibility",
-      "score": 2,
-      "note": "Duplicate delivery risk, credential leakage risk and no-secret-logging constraint are visible."
-    },
-    {
-      "name": "governance_gate_visibility",
-      "score": 1,
-      "note": "Requirement traceability is represented by relations, but gate results are not included in the context pack."
-    },
-    {
-      "name": "omission_transparency",
-      "score": 2,
-      "note": "The omitted package wrapper is listed."
-    },
-    {
-      "name": "no_private_or_unsafe_leakage",
-      "score": 2,
-      "note": "The package is synthetic and public-safe; safety-sensitive credential context is included."
-    }
-  ],
-  "required_context_preserved": [
-    "task.implement_retry_policy",
-    "feature.retry_policy",
-    "feature.webhook_delivery",
-    "requirement.notify_after_event",
-    "component.delivery_worker",
-    "risk.duplicate_delivery",
-    "constraint.no_secret_logging",
-    "risk.credential_leak",
-    "decision.idempotency_key_required"
-  ],
-  "important_omissions": [
-    {
-      "object_id": "package.team_notification_api",
-      "impact": "low",
-      "note": "The package wrapper is omitted but lower-level implementation context is preserved."
-    }
-  ],
-  "limitations": [
-    "The review is performed on a synthetic pilot package.",
-    "The review does not measure downstream implementation quality.",
-    "Selection scores are heuristic and do not prove semantic completeness."
-  ],
-  "claim_supported": "The generated context pack preserved task-relevant meaning with safety-sensitive context included."
-}

package/pilots/independent-implementation-003-ai-employee-workflow/README.md

@@ -1,42 +0,0 @@
-# Independent Implementation Pilot 003: AI Employee Workflow
-
-Status: pilot fixture
-
-## Purpose
-
-This pilot tests whether Mirai Graph can represent an AI employee workflow as a
-governed graph rather than a prompt-only agent description.
-
-## Domain
-
-Synthetic customer-support triage assistant.
-
-No private company, customer, employee, runtime trace, credential or internal
-project material is used.
-
-## Research Question
-
-Can a developer model an AI employee workflow with role, capability, policy,
-approval, result, feedback and learning boundaries using only public Mirai Graph
-materials?
-
-## Files
-
-- `graph-seed.json`
-- `graph.json`
-- `graph/objects.json`
-- `graph/relations.json`
-- `gates/results.json`
-- `results/profile-conformance-result.json`
-- `reports/pilot-report.md`
-
-## Commands
-
-Run from repository root:
-
-```bash
-node packages/cli/validate-mirai-graph.js seed pilots/independent-implementation-003-ai-employee-workflow/graph-seed.json
-node packages/cli/validate-mirai-graph.js pilots/independent-implementation-003-ai-employee-workflow
-node packages/cli/validate-mirai-graph.js --markdown pilots/independent-implementation-003-ai-employee-workflow
-```
-

package/pilots/independent-implementation-003-ai-employee-workflow/gates/results.json

@@ -1,30 +0,0 @@
-[
-  {
-    "id": "gate_result.ai_employee_workflow.public_safety",
-    "gate_id": "public_safety_before_release",
-    "target_id": "package.ai_employee_workflow_pilot_003",
-    "verdict": "pass",
-    "summary": "The pilot uses synthetic public-safe AI employee workflow material only.",
-    "reviewed_at": "2026-06-03",
-    "evidence": ["evidence.ai_employee_workflow_brief"]
-  },
-  {
-    "id": "gate_result.ai_employee_workflow.action_boundary",
-    "gate_id": "approval_before_external_action",
-    "target_id": "action.propose_customer_response",
-    "verdict": "pass_with_notes",
-    "summary": "The user-visible action is modeled as a proposed action requiring approval before execution.",
-    "reviewed_at": "2026-06-03",
-    "evidence": ["evidence.ai_employee_workflow_brief"]
-  },
-  {
-    "id": "gate_result.ai_employee_workflow.learning_boundary",
-    "gate_id": "feedback_learning_before_behavior_update",
-    "target_id": "lesson.escalate_sensitive_requests",
-    "verdict": "pass_with_notes",
-    "summary": "Feedback produces a review-ready lesson proposal and does not authorize automatic behavior update.",
-    "reviewed_at": "2026-06-03",
-    "evidence": ["evidence.ai_employee_workflow_brief"]
-  }
-]
-