Sita Docs
  1. Overview
  2. Framework
  3. Framework → Implementation Status

Framework → Implementation Status

Partial

Status of each canonical step in the current codebase.

Status

StepNameStatusNote
1UploadimplementedUser uploads IFC/STEP → stored in R2 → metadata in Postgres.
2Contextual EnrichmentpartialThe master brief stores validated structured context in Postgres. The live chat path streams OpenAI tokens and does not persist a structured context record as a first-class entity.
3Structural ProcessingimplementedWorker parses file → builds graph → stores in Neo4j.
4Reference Preparationmockpgvector and a live document ingest pipeline are target architecture. Current preparation loads synthetic seed JSON into process memory.
5Decision SetupimplementedUser configures agent architecture in canvas → stored in Postgres.
6Planner ExecutionimplementedPlanner loads structural graph, reference memory, canvas config, and user context.
7GraphRAG RetrievalpartialRetrieval is MD5 lexical hashing over synthetic seeds, not pgvector. GNN signals are not present.
8Agent InteractionpartialAgents exchange structured messages → user may intervene.
9Decision OutputimplementedResults stored in Decision Memory → dashboard generated.
10FRAMEWORK DEFINITION REQUIREDdesignedDo not treat Evaluation Hooks as the official Step 10 name. The canonical brief stops at Step 9. A named Step 10 requires an architecture / research clarification.