PPR Editor

A domain-aware workspace for modeling Product–Process–Resource structures and exporting them toward SysML v2 and AutomationML.

Live app →

Create, connect, move, inspect, validate, save, import, and export Products, Processes, and Resources. The workspace distinguishes definitions from usages, supports custom engineering attributes and semantic PPR relationships, does automatic layout, and ships coherent welding-cell, battery-module, and pump-assembly examples.

Stack

• Frontend — React, TypeScript, Vite, TanStack Query, React Flow
• Backend — Python 3.11, FastAPI, Pydantic v2
• Persistence — canonical model JSON
• API contract — OpenAPI + Orval codegen
• Agents — MCP server bridged to the canonical REST API

Pydantic is the canonical semantic model used by the API, persistence, validation, and exporters. The generic modeling kernel does not depend on PPR; PPR specializes it with domain types and relationship constraints. Standards exports are adapters over that canonical model.

Getting started

Requires Node.js 24, pnpm 10, Python 3.11, and uv.

pnpm install
uv sync

Run the API and the web workspace in two terminals:

PORT=8080 pnpm --filter @workspace/api-server run dev
PORT=23220 API_PORT=8080 pnpm --filter @workspace/ppr-modeler run dev

The workspace is then at http://127.0.0.1:23220, with the API proxied at /api. API_PORT tells the web dev server where the API is.

Tests

pnpm run test:ppr     # shared scenarios, backend, frontend, browser suites
pnpm run typecheck    # all packages
pnpm run release:smoke # read-only production-shaped API checks

Documentation

Document	Contents
docs/ppr-modeler-guide.md	User guide
docs/agent-integration.md	MCP client setup, tools, transports
docs/release-checklist.md	Release gate, smoke checks, recovery
replit.md	Architecture decisions and where things live

Where things live

• backend/model/core.py — generic modeling kernel
• backend/model/ppr.py — PPR specialization, relationship rules, validation
• backend/main.py — FastAPI service and JSON persistence
• backend/mcp_server.py — agent tools and resources over the REST API
• backend/sysml.py — typed SysML v2 projection and emitters
• backend/exporters.py — export adapters, including AutomationML
• lib/api-spec/openapi.yaml — API contract
• artifacts/ppr-modeler/ — graphical engineering workspace

Project status

This is a private, single-instance MVP, not a multi-tenant service:

• JSON files are the persistence boundary for the model and its checkpoints.
• Collaboration sessions are in memory and expire; they are not a durable store.
• There is no authentication, tenant isolation, or authorization. Access is controlled only by sharing a collaboration session ID and six-digit code.
• Desktop Chromium is release-tested; Firefox runs a portable smoke flow.

Do not deploy this as a public multi-user service without adding authentication, authorization, durable shared storage, and deployment isolation.

License

MIT