📒 Concepts
Everything in this section is learn-once reference. If you've already shipped a contract, you can skip ahead to the CLI Reference or Walkthroughs. If you're new, read these in order — each builds on the last.
In this section
| Page | What you'll learn |
|---|---|
| What is a Contract | The 5 required fields, the schema versioning model, and how Fluid Forge thinks about a "data product." |
| Builds, Exposes, Bindings | The three core blocks of every contract — how data is produced, what it exposes, and where it lands. |
| Providers vs Platforms | The difference between a binding.platform value and the provider plugin that handles it. Why local runs everywhere. |
| Quality, SLAs & Lineage | dq.rules, qos, and how lineage is auto-derived. Block bad deploys before they happen. |
| Governance & Policy | accessPolicy.grants, sovereignty constraints, and how Fluid Forge compiles to native cloud IAM. |
| Agent Policies (LLM/AI) | New in v0.7.1 — declarative boundaries on which AI models can read which fields. |
| vs alternatives | Honest comparisons against dbt, Terraform, Airflow, OPA, and Snowpark. |
In a hurry?
The single most important concept is the contract is the source of truth. Everything Fluid Forge does — schema validation, IAM compilation, SLA checks, lineage extraction — is derived from one YAML file you control. Everything else is mechanism.
What's next
- First time here? Read What is a Contract → Builds, Exposes, Bindings in order. ~15 minutes.
- Comparing tools? Skip to Forge vs dbt / Dagster / Terraform / Snowpark for the honest breakdown.
- Ready to build? Drop into Get Started and ship a real product locally in 30 seconds.