Data modeling and full-stack docs

Brick, 223P, SPARQL, CRUD APIs, Docker Compose, and lab automation for Open-FDD as a deployed platform now live in a separate repository:

open-fdd-afdd-stack — docs/

That site documents how the stack uses this open-fdd PyPI package under the hood (RuleRunner, YAML rules, pandas).

In this repository (rules engine only)

  • Column map resolvers — map Brick, Haystack, DBO, 223P, or vendor labels to DataFrame columns (dict, manifest, composite resolvers).
  • Expression rule cookbook — fault logic on pandas, including schedule and weather gates via params.schedule / params.weather_band.
  • examples/column_map_resolver_workshop/ — runnable ontology-agnostic demo (simple_ontology_demo.py).

The base PyPI wheel is the rules engine only. pip install "open-fdd[desktop]" adds the local FastAPI gateway, Feather-backed ingest, model.json on disk, and BRICK TTL generation under open_fdd.desktop (paths in open_fdd.desktop.storage.paths). Larger deployed platform concerns (Compose, production topology, extra RDF/SQL services) stay documented in open-fdd-afdd-stack.

AI-assisted modeling workflows

For the full OpenClaw + Codex OAuth + gateway HTTP integration picture, see Open FDD Claw architecture and scripts/OPENCLAW_RUNBOOK.md Phase 0.

For AI-assisted data modeling (OpenClaw, ChatGPT, or human-in-the-loop review), use a simple loop:

  1. Export model JSON from your backend (/model/export or stack export endpoint).
  2. Review and revise with an LLM (OpenClaw agent or ChatGPT web UI).
  3. Validate the edited JSON before import.
  4. Import JSON back to backend (/model/import) and re-run SPARQL/rules checks.

The same flow works for:

  • OpenClaw agents running local automation loops.
  • ChatGPT online interface where a human copies export JSON in and validated JSON out.
  • Hybrid workflows where AI drafts and human confirms before import.

For robust prompts, import schema guidance, and operator-safe pre-flight checks, see:

Local HTTP gateway note (open-fdd repo)

The FastAPI gateway (open_fdd.gateway, CLI open-fdd-gateway / open-fdd-desktop-bridge) supports agent-friendly backend operations such as:

  • model export/import/validate,
  • SPARQL query endpoints (/data-model/sparql, /data-model/sparql/upload),
  • timeseries bounds/query over Feather data,
  • weather/BACnet ingest and ML training routes.

This enables OpenClaw-style local assistants to do data modeling, retrieve and join data in pandas/Feather workflows, run faults, and iterate with a human operator. For how to run the gateway locally, storage paths, and start-local, see Desktop app.