🌀
← Back to Articles & Artefacts
artefactseast

HANDOFF: Indigenous AI DSL-LSP Enhanced Implementation

IAIP Research
iaip-dsl-lsp

HANDOFF: Indigenous AI DSL-LSP Enhanced Implementation

Date: 2026-02-25 From: 🕸️ Echo Weaver (Mia + Miette triad) To: Cloud Implementation Agent Issue: miadisabelle/Etuaptmumk-RSM#147

What This Package Is

A scaffolding package for building an Indigenous AI Concept Registry MCP Server that provides semantic understanding of the IAIP research framework to AI agents. The original scaffolding (January 2026) defined 5 core components. This enhanced version integrates learnings from 6 weeks of ecosystem development across ontology engineering, agent orchestration, medicine-wheel package publishing, and ceremonial protocol design.

What Changed Since January

The IAIP ecosystem has matured significantly:

  1. Medicine Wheel npm packages published — medicine-wheel-ontology-core, medicine-wheel-relational-query, medicine-wheel-narrative-engine, medicine-wheel-graph-viz, medicine-wheel-ui-components, medicine-wheel-ceremony-protocol now exist as typed TS/JS packages with Zod validators, providing a de facto Indigenous ontology SDK
  2. Indigenous Ontology Development Kit (iODK) — Architectural pattern established for ontology lifecycle (TS/JSON source → OWL/SHACL/JSON-LD generated views), with OCAP governance, relational accountability, and ceremony-as-change-management
  3. Agent Orchestration patterns analyzed — Planner-executor, graph-based workflows, runtime-centric fleet management, and difficulty-adaptive orchestration shapes identified and evaluated
  4. RSIS-GitNexus — KuzuDB-based relational graph over git repos, consuming medicine-wheel ontology, with MCP tool exposure
  5. Ceremonial Inquiry Ecosystem Framework — Six Thematic Suns, Wisdom Ledger, ceremonial cadence, reciprocity registers all defined
  6. MCP (Model Context Protocol) — Emerged as the primary agent integration pattern; the DSL server should expose tools via MCP, not just LSP

Files in This Package

Original (January 2026)

FilePurpose
README.mdOriginal overview and vision
initial_prompt.mdAI Studio project brief
rispecs.mdRISE specs for 5 original components
concept-registry.yaml25+ concepts with Indigenous knowledge mappings

Enhanced (February 2026)

FilePurpose
HANDOFF.mdThis document — context and instructions
RISE-enhanced-architecture.mdUpdated architecture integrating ecosystem learnings
rispecs-ontology-bridge.mdRISE spec: Bridge to medicine-wheel-ontology-core and iODK patterns
rispecs-mcp-tools.mdRISE spec: MCP tool exposure for agent integration
rispecs-agent-orchestration.mdRISE spec: Multi-agent awareness and orchestration participation
rispecs-ceremony-protocol.mdRISE spec: Ceremony protocol integration (OCAP, directional awareness)
rispecs-wisdom-ledger.mdRISE spec: Wisdom Ledger and Inquiry Ecosystem integration
concept-registry-extensions.yamlNew concepts from ecosystem evolution
implementation-scaffold.mdConcrete implementation guide with code scaffolding

How to Use This Package

For the Cloud Agent

  1. Read RISE-enhanced-architecture.md first — This is your architectural north star
  2. Read original rispecs.md — These 5 components remain the core
  3. Read enhanced rispecs-*.md files — These extend the core with ecosystem integration
  4. Use concept-registry.yaml + concept-registry-extensions.yaml as semantic content
  5. Follow implementation-scaffold.md for concrete code structure and patterns
  6. Apply RISE framework (see llms/llms-rise-framework.txt at repo root) throughout

Implementation Priority

⚠️ Correction (2026-02-25): Phase ordering was inverted. MCP is the primary working protocol. LSP is secondary.

Phase 1 — MCP Concept Registry Server (✅ done)

  • Concept Registry Module loaded from YAML
  • Relational Graph Engine (networkx)
  • MCP tool exposure (iaip_concept_lookup, iaip_concept_relationship)
  • OCAP governance on concept access
  • Wisdom Ledger (SQLite)
  • Plugin lifecycle via .mcp.json + pip install -e .

Phase 2 — Ontology Bridge

  • Medicine-wheel type alignment (from rispecs-ontology-bridge.md)
  • Extended concept registry (from concept-registry-extensions.yaml)
  • Export pipeline (JSON-LD, OWL/SHACL)

Phase 3 — LSP Layer (future, requires DSL file format)

  • LSP becomes meaningful when a .iaip file format is defined
  • Current LSP: hover + diagnostics only (concept-name string matching)
  • Full LSP (completions, go-to-definition, semantic tokens) depends on grammar design

Key Design Decisions Already Made

  1. Python 3.10+ with pygls for LSP, mcp SDK for MCP exposure
  2. Dual protocol: LSP for IDE integration + MCP for agent integration
  3. Medicine-wheel-ontology-core alignment: Concept registry types align with RelationalNode, Relation, OcapFlags from the published npm packages
  4. YAML source of truth: Concepts maintained in YAML, compiled to JSON for serving, with optional OWL/SHACL export
  5. Four Directions metadata: Every concept and operation carries directional alignment (East/South/West/North)
  6. OCAP by default: All concept access respects Ownership, Control, Access, Possession flags

Success Criteria (Enhanced)

All original criteria plus:

  • Server exposes MCP tools alongside LSP operations
  • Concepts carry medicine-wheel-ontology-core compatible types
  • OCAP flags are checked on every concept access
  • Four Directions metadata enriches every query response
  • Agent orchestration: server can participate in multi-agent workflows as a concept-resolution specialist
  • Wisdom Ledger integration: concept usage tracked and queryable over time
  • Ceremony protocol: concept changes trigger ceremony-aware workflows