open-fdd-modeling

name	open-fdd-modeling
description	AI-assisted Open-FDD data model export, LLM review, validate, import, and SPARQL checks via the desktop bridge.

Open-FDD data modeling (OpenClaw)

When to use

• Operator wants to adjust sites, equipment, or points with help from the assistant.
• You have (or can fetch) model JSON from the Open-FDD bridge and need a safe edit → validate → import loop.

Preconditions

• Bridge running (default http://127.0.0.1:8765; from Docker use http://host.docker.internal:8765 per runbook).
• Use GET /docs or /openapi.json for exact request bodies on this revision.

Workflow

1. Export — GET {bridge}/model/export — capture current JSON.
2. Draft changes — Propose edits only in structured JSON the bridge accepts; prefer minimal diffs and preserve site_id / UUIDs unless the operator asks to create new entities.
3. Validate — POST {bridge}/model/validate with the draft payload when available; otherwise use import validation errors from POST /model/import response.
4. Import — POST {bridge}/model/import only after operator confirmation.
5. Graph check — Optional POST {bridge}/data-model/sparql with a small SELECT to sanity-check Brick/TTL expectations.

Safety

• Never invent API keys (BACnet DIY, OpenClaw gateway token, etc.); read from operator env or bridge config endpoints if documented.
• On failure, return HTTP status, response body snippet (redacted), and the curl the operator can replay.

References

• Repository: docs/modeling/index.md, docs/open-fdd-claw-architecture.md, scripts/OPENCLAW_RUNBOOK.md.