Ask
Questions in plain English, answered from the live mirror.
Tables
What you can query, and what each column holds.
Pick a table to see its columns.
SQL
Read-only. SELECT and WITH statements only.
Saved queries
Shared with everyone who has access to the tables involved.
Sources
The databases this server reads from. Nothing here is ever written to.
Sync jobs
Each job pulls its own extraction from one source into the mirror.
Recent runs
Schema training
What you write here is what the model reads before every query.
Pick a table to describe it.
Relationships
How tables join. One per line, in plain words.
Business rules
Definitions and filters the model should always apply.
Context preview
Exactly what gets sent to the model.
Connect
Everything you need to point Claude, or any other client, at this server.
Access tokens
People
Who can sign in, and which tables they see.
Models
Local models served by Ollama on this machine.
Good CPU-only choices
- qwen2.5:3b-instruct-q4_K_M — ~2.0 GB, strong at SQL, the default
- qwen2.5-coder:7b-instruct-q4_K_M — ~4.7 GB, better SQL, needs ~8 GB RAM free
- llama3.2:3b-instruct-q4_K_M — ~2.0 GB, better prose, weaker SQL
- phi3.5:3.8b-mini-instruct-q4_K_M — ~2.2 GB, fast on few cores
Activity
Every query run through this server.