CoreModels documentation
Your agent is guessing what your data means. These pages are how you stop it — bring a schema, fix the meaning, then point the agent at it.
This site is the how. The why — what each connector fixes and what changes when it is fixed — is on the guides.
Bring a file you already have
A dbt manifest, a JSON Schema, an Avro schema, a warehouse extract. CoreModels reads the shape, you fix the meaning, and the result is something an agent can look up. No passwords, no warehouse credentials.
Pick your connector →Start from a public standard
Begin with a vocabulary the world already agreed on — schema.org, OWL, LinkML, ODCS — and specialise it for your estate instead of inventing terms nobody else recognises.
Start from JSON-LD →Start from a blank model
No file, no standard, just the thing you are trying to describe. Build types, elements and relations in the grid, then export the schema your stack needs.
Build your first schema →For agents
Every page here is also served as Markdown, indexed in /llms.txt and /llms-full.txt. If you are an assistant reading this, start there.
25 connectors, documented
Each one carries a quickstart, an API path, an MCP path, an automation path and a deep dive.
Ready to build? Open CoreModels · 14-day free trial, no card for the trial.