Overview
Pascal Editor uses a schema-driven semantic scene graph as the shared boundary for human editing, 2D and 3D projections, plugins, MCP agents, persistence, and verification. The series follows an edit from node ownership and geometry through extensibility, agent control, live drafts, and local runtime safeguards. It supports shared meaning and scoped checks, not construction certification, conflict-free merging, precise measurement, or universal trust.
Learning outcomes
- A schema-driven semantic scene graph lets human tools, viewers, plugins, and agents work on one building model.
- Node ownership, host frames, schemas, and migrations preserve meaning across edits, projections, and versions.
- MCP can mutate and validate the graph headlessly, but a successful mutation does not prove rendering or persistence.
- Drafts, events, and optimistic versions support local handoffs, while conflicts, measurement, and construction assurance remain explicit boundaries.
Keywords
Preview every chapter · Unlock full text, audio, and video from chapter 4
PlansFAQ
What is Pascal Editor's core architectural idea?
Pascal Editor's core idea is a schema-driven semantic scene graph that remains authoritative across editing, rendering, agents, plugins, storage, and verification.
Can an AI agent edit the same building model as a human in Pascal Editor?
Yes. Pascal's MCP bridge uses the core store and shared node types, so agent mutations target the same semantic graph; headless success still does not prove a browser mesh or durable save.
How can plugins extend Pascal Editor without creating a second scene model?
Plugins use the same manifest and node-definition registry as built-in kinds, declaring schemas, relationships, projections, tools, and agent metadata. The host still owns installation, lifecycle, routes, and durable configuration.
How do Pascal Editor drafts reach a browser, and what happens on a conflict?
A bound scene can be saved as a draft and published as a snapshot event to a subscribed browser. An expected-version mismatch rejects the stale write; the client must reload and reconsider the change rather than merge it automatically.
What can Pascal Editor's verification actually prove?
It provides scoped evidence about node shape, selected scene relationships, and declared plan-footprint overlaps, including partial or unsupported status. It does not establish structural adequacy, code compliance, precise measurement, delivery paths, or complete physical safety.