estatewise-local-stack - SKILL.md Agent Skill

name: estatewise-local-stack description: Manual runbook for local setup, service combinations, and environment debugging in EstateWise. Use when you need to start or troubleshoot the local stack. argument-hint: [service-or-scenario] disable-model-invocation: true

EstateWise Local Stack

Invoke this skill manually when the task is about running or debugging the local environment.

Pick The Smallest Service Combination

Frontend + backend only

Use for chat, auth, conversation, forums, and most UI work:

# repo root
npm run dev

Or individually:

cd backend && npm start
cd frontend && npm run dev

Backend only

Use for API, auth, Swagger, metrics, graph, or conversation work:

cd backend
npm start

Useful endpoints:

http://localhost:3001/api-docs
http://localhost:3001/swagger.json
http://localhost:3001/metrics
http://localhost:3001/status

Frontend only

Use when changing local tRPC pages or static UI, or when backend work is not required:

cd frontend
npm run dev

Note: frontend local tRPC endpoints live inside Next.js and do not require the Express backend.

MCP only

Use when adding or debugging tools:

cd mcp
npm install
npm run build
npm run client:dev

npm run dev waits for a client and appears idle. That is expected.

Agentic AI + MCP

Use for orchestration or tool-consumer debugging:

cd mcp && npm install && npm run build
cd ../agentic-ai && npm install && npm run build
npm run dev "Find 3-bed homes in Chapel Hill and estimate mortgage"

gRPC

Use for market pulse service work:

cd grpc
npm install
npm run dev

Deployment Control

Use for deployment dashboard/API work:

cd deployment-control
npm run install:all
npm run dev
npm run dev:ui

Environment Checklist

Common backend env:

MONGO_URI
JWT_SECRET
GOOGLE_AI_API_KEY
PINECONE_API_KEY
PINECONE_INDEX

Graph-specific env:

NEO4J_ENABLE=true
NEO4J_URI
NEO4J_USERNAME
NEO4J_PASSWORD
NEO4J_DATABASE

MCP-specific env:

API_BASE_URL
FRONTEND_BASE_URL
A2A_BASE_URL
MCP_TOKEN_SECRET

High-Frequency Local Debugging Problems

Frontend hits deployed backend unexpectedly

Search for hardcoded URLs:

rg -n "estatewise-backend\\.vercel\\.app|API_BASE_URL" frontend

Do not assume frontend/lib/api.ts is the only call site.

MCP seems hung

It is probably waiting for a stdio client. Use:

cd mcp && npm run client:dev

Graph endpoints return 503 or empty data

Check Neo4j env and whether graph ingest actually ran.

Agentic AI fails to call tools

Confirm MCP built successfully and that tool names in the consumer runtime still match the MCP registry.

Deployment-control works locally but is operationally unsafe

Remember it has no built-in auth/RBAC. Only run against trusted environments.

Validation Reference

Backend: cd backend && npm run build && npm run test
Frontend: cd frontend && npm run build && npm run test
MCP: cd mcp && npm run build && npm run client:call -- <tool> '<json>'
Agentic AI: cd agentic-ai && npm run build && npm run dev "goal"
gRPC: cd grpc && npm run build && npm run test
Deployment Control: cd deployment-control && npm run build