SSpeckle

iDBQuery's Speckle connector links a project from app.speckle.systems (or any self-hosted Speckle server) into the chat. Paste the project URL, paste a Personal Access Token, pick one or more models, and the latest version's objects sync into a per-source SQLite — flattened with hot-path columns for ifc_type, level, area, volume, and the full Property Sets in JSON. The split chat layout puts the AI on the left and the live 3D viewer on the right; ask 'highlight all windows' and the model lights up via Speckle's native filter, or 'walls > 50 m²' and the AI runs SQL → flashes the matching set in 3D. Click any object → 'Ask about it' chip pre-fills a question with full metadata in the answer.

• Personal Access Token auth (paste & go — no OAuth dance)
• Connect by URL — paste any project / model link
• Multi-model per source — sync architecture, structure, MEP together
• Object-level sync into SQLite (speckle_objects table)
• Hot-path columns: ifc_type, level, area, volume, units, application_id
• Property Sets queryable via json_extract on properties_json
• Embedded 3D viewer alongside the chat — drag to resize
• LLM-driven highlight: native property filter (instant) + ad-hoc id-based
• Group coloring: 'color rooms by area', 'tag load-bearing walls red'
• Click an object in the viewer → 'Ask about it' chip in the chat
• Auto-sync via Speckle webhooks (opt-in) + nightly safety cron
• Self-hosted Speckle servers supported
• speckle_view widget for reports — pinned 3D in your dashboard