Warnerco Schematica

WARNERCO Robotics Schematica

Agentic robot schematics system with semantic memory and retrieval-augmented generation.

Architecture

┌─────────────────────────────────────────────────────────────┐
│                     FastAPI + FastMCP                       │
├─────────────────────────────────────────────────────────────┤
│  LangGraph Flow (7-node Hybrid RAG)                         │
│  parse_intent -> query_graph -> inject_scratchpad -> retrieve│
│  -> compress -> reason -> respond                            │
├─────────────────────────────────────────────────────────────┤
│  Hybrid Memory Layer                                        │
│  +-------------------+  +-------------------+  +-----------+│
│  | Vector Store      |  | Graph Store       |  | Scratchpad|│
│  | JSON->Chroma->    |  | SQLite + NetworkX |  | In-memory |│
│  | Azure AI Search   |  | (Knowledge Graph) |  | (Session) |│
│  +-------------------+  +-------------------+  +-----------+│
└─────────────────────────────────────────────────────────────┘

Project Structure

src/warnerco/backend/
├── app/
│   ├── main.py           # FastAPI application
│   ├── config.py         # Settings and environment
│   ├── models.py         # Pydantic schemas
│   ├── routes.py         # API endpoints
│   ├── mcp_tools.py      # FastMCP tool definitions
│   ├── adapters/         # Memory backend implementations
│   │   ├── json_store.py
│   │   ├── chroma_store.py
│   │   ├── azure_search_store.py
│   │   ├── graph_store.py
│   │   └── scratchpad_store.py
│   └── langgraph/
│       └── flow.py       # 7-node hybrid RAG orchestration
├── data/
│   ├── schematics/       # JSON source of truth
│   └── chroma/           # Vector embeddings
├── static/dash/          # SPA dashboards
└── .env                  # Configuration

Commands

cd src/warnerco/backend

# Local development
uv sync
uv run uvicorn app.main:app --reload --port 8000

# Index schematics into Chroma
uv run python -c "from app.adapters.chroma_store import ChromaMemoryStore; import asyncio; asyncio.run(ChromaMemoryStore().index_all())"

# MCP stdio server (for Claude Desktop)
uv run warnerco-mcp

Memory Backend Selection

Set MEMORY_BACKEND in .env:

MCP Tools

LangGraph Flow

7-node hybrid retrieval-augmented generation:

1. parse_intent - Classify query (lookup/diagnostic/analytics/search)
2. query_graph - Enrich with knowledge graph relationships
3. inject_scratchpad - Add session working memory
4. retrieve - Fetch candidates from memory backend
5. compress_context - Minimize token bloat
6. reason - LLM generates response (Azure OpenAI gpt-4o-mini)
7. respond - Format for dashboards/MCP

Adding Schematics

Edit data/schematics/schematics.json:

{
  "id": "WRN-00026",
  "model": "WC-900",
  "name": "New Robot Name",
  "component": "component description",
  "version": "v1.0",
  "summary": "Technical summary...",
  "category": "sensors",
  "status": "active",
  "tags": ["tag1", "tag2"],
  "specifications": {
    "spec_key": "spec_value"
  },
  "url": "https://schematics.warnerco.io/..."
}

Then re-index: uv run python -c "...index_all()"

Dashboards

• Schematics Browser (/dash/schematics/) - Search, filter, view robot data
• Memory Learning (/dash/memory/) - Educational RAG visualization

Azure Deployment

See references/azure-deployment.md for:

• Container App setup
• APIM configuration
• AI Search indexing
• OpenAI model deployment

API Endpoints

Converted and distributed by TomeVault — claim your Tome and manage your conversions.