POST /api/graph/ontology/generate
Extract the entity types, attributes, and relationship types that fit the question.
Move from source material, relationship graphs, and agent personas to dual-platform simulation, reporting, and follow-up questions. This is a complete workflow using fictional public-style sample data.
5-stage workflow · Visual walkthroughExtract ontology, facts, and relationships from the source material to provide context for later agent generation.
POST /api/graph/ontology/generate
Extract the entity types, attributes, and relationship types that fit the question.
POST /api/graph/build
Chunk the material, extract entities and relationships, and inject facts and collective memory.
After the graph is ready, create a simulation instance and enter Environment Setup.
Open a few cards to spot-check the source entity, interests, stance, and description; confirm unrelated entities did not enter this simulation.
Return to the material and variables. Check whether personas are reasonable, logs are meaningful, and the report has evidence behind it. Then iterate with a one-variable comparison.
Demo statusThis is an offline UI reconstruction using fictional sample data. It does not upload files, call APIs, or run a real simulation. The provided screenshot is used only to recreate the visual layout of Environment Setup.
The player supports dragging, play/pause, restart, chapter jumps, 0.75×–2× playback speed, and keyboard control (Space, ←/→, Home/End). It is 150 seconds long and covers all five upstream stages.
Read the verified user guide →Steps are grounded in the official repository README, environment-variable example, and current frontend component structure. Public videos were used only to verify the workflow, not copied.