| Plan | Dependencies | Needed By |
|---|---|---|
| NxWorkspaceAndBoundaries | None (foundational) | All other plans |
| Walking Skeleton | Nx Workspace | Validates stack viability |
| SEA-DSL Toolchain | Nx Workspace | Compiler Chain, Semantic Core |
| Compiler Chain | SEA-DSL Toolchain | Most implementation plans |
| Knowledge Graph Runtime | Compiler Chain, Nx Workspace | Semantic Core, Cognitive |
| Semantic Core Services | Knowledge Graph, SEA-DSL Toolchain | LLM Provider, Cognitive |
| Runtime Kernel | Nx Workspace, Compiler Chain | Protocol Gateway, Messaging |
| Protocol Gateway | Runtime Kernel | Platform UI |
| Messaging Substrate | Runtime Kernel | Integration needs |
| Semantic Identity | Messaging | Governance, Observability |
| GovernedSpeed™ Runtime | Messaging, Identity | LLM Provider, Agent Society |
| Observability | GovernedSpeed™, Identity | All runtime services |
| Temporal Memory | Observability, Knowledge Graph | Cognitive, Simulation |
| LLM Provider | GovernedSpeed™, Observability | Cognitive, Agent Society |
| Simulation & Replay | Temporal Memory, Observability | TBD |
| Agent Society Runtime | LLM Provider, GovernedSpeed™ | Cognitive Extension |
| Cognitive Extension | LLM Provider, Agent Society, Semantic Core | PET, Case Mgmt, UI |
| Workbench UI | Protocol Gateway, Cognitive | User-facing |
| Platform UI Integration | Protocol Gateway, Cognitive | User-facing |
| Semantic Debt | GovernedSpeed™, Identity | Governance flows |
| Documentation System | Compiler Chain, Cognitive | Developer experience |
| PET Training App | Cognitive Extension | User-facing |
| MECE Quadrant | Cognitive Extension, GovernedSpeed™ | Advanced features |
| Case Management | Cognitive Extension | Business logic |
| User Feedback Loop | Cognitive Extension | Learning |
| VibesPro™ Integration | Cognitive, GovernedSpeed™ | Advanced features |
| # | Plan Name | Key Dependencies | Reasoning |
|---|---|---|---|
| 1 | NxWorkspaceAndBoundaries |
None | Foundation layer. Establishes Nx workspace structure, module boundary enforcement (ESLint rules), and the “kernel-shell” topology per ADR-033/034. Without this, other plans have no framework for code organization or boundary enforcement. |
| 2 | WalkingSkeleton |
#1 | Stack validation. Proves end-to-end connectivity of all major components (SEA-DSL → LLM → Vector/Graph → Policy) before building features. Must run early to catch integration issues. |
| 3 | SeaDslToolchain |
#1 | Language foundation. Implements SEA-DSL parsing, validation, and formatting (tree-sitter grammar, OPA policy evaluation). Required before any semantic compilation can occur. |
| 4 | CompilerChain |
#3 | Delivery pipeline. Implements the 7-stage compiler chain (Specs → Plan → SEA™ DSL → AST → IR → Manifest → Generated Code). This is the core “no-agent-writes-code” invariant enforcement mechanism. |
| 5 | KnowledgeGraphRuntime |
#1, #4 | Semantic storage. Oxigraph RDF store + SPARQL/SHACL support. Needed by Semantic Core and Cognitive services for semantic context queries. |
| 6 | RuntimeKernel |
#1, #4 | Execution core. Port-driven kernel with command/query buses, unit-of-work, and event publication. The kernel is independent of shells (APIs/CLIs). |
| 7 | ProtocolGateway |
#6 | Inbound adapters. HTTP API, OpenAI-compatible facade, LSP endpoints. These are shells that translate protocols to kernel commands. Depends on kernel being defined. |
| 8 | MessagingSubstrate |
#6 | Cross-context communication. NATS JetStream with outbox/inbox patterns. Required for reliable event publishing between bounded contexts. Supersedes the old IntegrationGateway. |
| 9 | SemanticIdentity |
#8 | Identity & provenance. IFL (Internal Federated Ledger), ifl:hash/ifl:token dual-state identity, Merkle proofs. Needed before governance decisions can be attestable. |
| 10 | GovernedSpeed™Runtime |
#8, #9 | Unified governance. Consolidates Policy Gateway + Risk/Evidence + CALM validation. Depends on messaging and identity for audit trails. |
| 11 | ObservabilityAndEvidence |
#9, #10 | Telemetry layer. OpenTelemetry + OTel Collector + semantic correlation. Depends on governance runtime for evidence capture and identity for provenance. |
| 12 | SemanticCoreServices |
#3, #5 | Core services. SEA-DSL registry, policy evaluation, and KGS integration. Builds on toolchain and knowledge graph. |
| 13 | TemporalMemoryAndPatternOracle |
#5, #11 | Institutional memory. pgvector + TimescaleDB + Pattern Oracle. Depends on KG for semantic context and observability for metric ingestion. |
| 14 | LlmProviderAbstraction |
#10, #11 | AI infrastructure. LiteLLM unified interface. Explicitly depends on GovernedSpeed™ (LLM calls route through Policy Gateway) and Observability (spans). |
| 15 | AgentSocietyRuntime |
#10, #14 | Multi-agent platform. Agent registry, tool execution, HITL escalations. Needs governance and LLM provider to be operational. |
| 16 | CognitiveExtensionLayer |
#12, #14, #15 | AI services. Context analysis, artifact generation, CADSL renderer. Depends on Semantic Core, LLM provider, and agent infrastructure. |
| 17 | SimulationAndReplay |
#11, #13 | What-if scenarios. Deterministic simulation from snapshots. Depends on observability traces and temporal memory for pattern comparison. |
| 18 | SemanticDebtAndIncidents |
#9, #10 | Governance workflows. Debt ledger, incident grouping, change proposals. Needs identity for audit trails and governance for enforcement. |
| 19 | DocumentationSystem |
#4, #16 | Automated docs. Generates docs from specs and manifests. Depends on compiler chain for traceability and cognitive layer for context analysis. |
| 20 | PlatformUIIntegration |
#7, #16 | Shell integration. Zed IDE + LibreChat shells with LSP/OpenAI endpoints. Depends on protocol gateway and cognitive services. |
| 21 | WorkbenchUI |
#7, #16 | Operator console. KGS explorer, manifest inspector, governance console. Depends on protocol gateway and cognitive for rendering. |
| 22 | PetTrainingApp |
#16 | User-facing app. Prompt training with gamified learning. Depends on cognitive extension for the judge pipeline. |
| 23 | CaseManagement |
#16 | Business logic. CMMN-based case management with artifact pipelines. Depends on cognitive extension for task automation. |
| 24 | UserFeedbackLoop |
#16 | Learning integration. Feedback capture, RL signals, personalization. Depends on cognitive extension for adaptive responses. |
| 25 | MeceQuadrantIntegration |
#10, #16 | Advanced architecture. Soul/Mind/Will/Body quadrant orchestration. Advanced feature depending on governance and cognitive layers. |
| 26 | VibesPro™Integration |
#10, #16 | External integration. VibesPro™ substrate (Temporal AI, type sync, meta-generators). Advanced feature integrating governance and cognitive. |
The ordering follows these principles:
Foundation First (1-4): Workspace structure → Stack validation → Language tooling → Compiler pipeline. These are non-negotiable prerequisites.
Infrastructure Layer (5-11): Knowledge storage → Kernel → Gateways → Messaging → Identity → Governance → Observability. Each layer builds on the previous.
Core Services (12-16): Semantic Core → Temporal Memory → LLM → Agents → Cognitive. These implement the business logic atop the infrastructure.
Advanced Features (17-26): Simulation, Debt/Incidents, Documentation, UIs, Apps, and integrations. These are higher-level features that consume the core services.