Basic TODOs:

• Fix the value conversion in migration modules

Layer 1 - Human feedback loop

Starting the app

• Communicate the current settings setup, what seems to be missing
• Enable LLM-powered Input if something is missing to fix the config

Analyze SQL tables

• Present the status of tables, provide info what tables are interesting and why, what are skipped and why (enable input)
• Present the user each table, and how it is being ported to Memgraph (enable input)
• Present the user how each index is being ported into the Memgraph (enable input)

Migrations

• Give to user the queries (migration modules) used for each table
• Validate the migration and present to user the status compared to SQL

Layer 2 - SQL <> Memgraph API

Layer 3 - Adaptability

Tehnical notes for @antejavor

Do I really need a LangGraph?

How could I leverage https://www.tensorzero.com/ for better prompting