Architecture Documentation Migration Plan
This audit classifies the current architecture documentation against the current AncientOS kernel model. It is a planning and traceability document for the documentation pass; it does not add runtime behavior.
Classification
| Document | Classification | Reason |
|---|---|---|
docs/architecture.md |
substantial rewrite | It was historically accumulative and described AncientOS as a continuity layer/app runtime instead of the governed cognition kernel. |
docs/canonical_terminology.md |
substantial rewrite | It classified Keeper as an app/workflow and used substrate/app language that blurred kernel services. |
docs/app_component_boundary_model.md |
substantial rewrite | It needed replacement framing around kernel services, applications, providers, and transports. |
docs/architecture/kernel_services.md |
new canonical support page | No concise role map existed for Keeper, Beastmaster, Lich, Zeus, execution lanes, and provider selection. |
docs/architecture/keeper.md |
substantial rewrite | Keeper needed correction from task system behind Luna to objective/task kernel service. |
docs/rubick_capability_registry.md |
minor edits | Rubick was mostly accurate but needed repository/knowledge graph and kernel role language. |
docs/oracle_visibility.md |
minor edits | Oracle needed kernel evidence synthesis language instead of Luna standalone service framing. |
docs/memory_architecture.md |
minor edits | LifeVault was accurate but needed explicit kernel-service wording. |
docs/inference_doctrine.md |
minor edits | It needed current local-first/frontier escalation posture and model-mode language. |
docs/operational_cognition_routing.md |
minor edits | It needed runtime-kernel framing rather than Luna-owned routing. |
docs/terminal_transport.md |
minor edits | It needed the current host-run TUI URL override behavior. |
docs/architecture/rubick_repository_intelligence_and_handoff.md |
minor edits | It was current but needed kernel capability wording. |
docs/architecture/rubick_knowledge_graph.md |
minor edits | It was current but needed kernel relationship wording. |
docs/architecture/governance_kernel.md |
no broad change | It is already the canonical lifecycle authority document. |
docs/architecture/keeper_calendar_boundary.md |
no broad change | It already treats Keeper calendar behavior as a governed boundary. |
docs/architecture/keeper_handoff.md |
no broad change | It already describes read-only, evidence-preserving handoff behavior. |
docs/architecture/keeper_import_boundary.md |
no broad change | It already preserves import boundaries and approval constraints. |
docs/architecture/keeper_import_proposal.md |
no broad change | It is a bounded Keeper import proposal rather than canonical kernel taxonomy. |
docs/architecture/keeper_import_review.md |
no broad change | It remains a review artifact and does not redefine architecture. |
docs/architecture/keeper_inbox_triage.md |
no broad change | It already frames triage as planning support, not execution authority. |
docs/architecture/keeper_lifecycle.md |
no broad change | It already describes objective state without changing approval authority. |
docs/architecture/keeper_objective_instantiation.md |
no broad change | It already preserves objective instantiation as governed planning. |
docs/architecture/keeper_operational_planning_doctrine.md |
no broad change | It already keeps planning separate from execution and approval. |
docs/architecture/keeper_scheduling.md |
no broad change | It already treats scheduling as bounded Keeper planning behavior. |
docs/architecture/lich_intake_gate.md |
no broad change | It already describes Lich as the approval gate. |
docs/architecture/loop_layer.md |
no broad change | It already keeps loop behavior observational and governed. |
docs/architecture/objective_intake.md |
no broad change | It already defines governed intake without transport-specific runtime assumptions. |
docs/architecture/objective_intake_resolver.md |
no broad change | It already preserves resolver behavior as read-only classification/planning support. |
docs/architecture/ser6buzz_maintenance_dashboard.md |
no broad change | It is an app-specific architecture note and does not redefine kernel services. |
docs/architecture/strategy_selection.md |
no broad change | It is app/objective strategy documentation and remains outside this canonical kernel pass. |
Proposal documents under docs/proposals/ |
leave as historical debt | They are proposal/rationale records. Do not rewrite them as current doctrine in this pass. |
Generated docs under docs/api/, docs/_build/, and site/ |
no manual edits | Generated output should only change through the docs build workflow. |
Dependency Order
- Establish kernel vocabulary in
architecture.mdandcanonical_terminology.md. - Correct application/provider/transport boundaries in
app_component_boundary_model.md. - Add the compact kernel service responsibility map.
- Correct role-specific pages: Keeper, Rubick, Oracle, LifeVault, inference, runtime routing, terminal transport, repository reasoning, and knowledge graph.
- Leave historical proposals and roadmap debt explicitly identified instead of mixing historical evolution with current doctrine.
Remaining Debt
docs/luna_master_roadmap.mdis still historically accumulative and should eventually be split into current architecture, roadmap, and historical notes.- Proposal documents that have been adopted should gain a standard historical banner pointing to canonical docs.
- Knowledge surface files under
knowledge/may still carry older classifications and should be audited separately. - Machine-readable manifests may need a later schema pass to reflect
ancientos_kernel_serviceclassifications consistently.