Skip to main content
canonic setup
Bootstraps a new canonic project: name → first connection (connection-tested before proceeding) → LLM → optional schema preview → writes canonic.yaml plus scaffold directories and a .gitignore. Progress is checkpointed to .canonic/setup-state.json, so an interrupted run resumes where it left off. Run inside an already-set-up project, it offers a status/add-connection menu instead of overwriting committed files.

​
What happens after the config is written

The wizard continues straight into the golden path:
1

Bootstrap the first connection

Tier-1 schema introspection produces deterministic semantic sources.
2

Run a first answer

A demo metric query returns result rows plus the metadata band (freshness, guardrails, final/provisional).
3

Hand off

You’re shown what to review, the exact query call that was just run, and how to connect an agent over MCP.
This is also the command behind bare canonic with no subcommand and no --json; see CLI Overview.