Airlayer - In-Process Semantic Engine

airlayer

In-process semantic engine. Compiles YAML semantic definitions to dialect-specific SQL. Designed as a tool-use interface for LLMs.

GitHubApache-2.0

Semantic layer

Define views with dimensions and measures. Query them from the CLI. Airlayer compiles to dialect-specific SQL.

views/orders.view.yml

name: orders
table: public.orders
dialect: postgres

dimensions:
  - name: status
    type: string
    expr: status
  - name: created_at
    type: time
    expr: created_at

measures:
  - name: total_revenue
    type: sum
    expr: amount
  - name: order_count
    type: count
    expr: id

Install

$ bash <(curl -sSfL https://raw.githubusercontent.com/oxy-hq/airlayer/main/install_airlayer.sh)

Installs the airlayer binary to your system.

Supported databases

postgressnowflakebigqueryduckdbmotherduckmysqlclickhousedatabricksredshiftsqlite

Quick start

Initialize airlayer in your project directory. The CLI will walk you through database connection and schema discovery.

terminal

$ airlayer init

This generates:

config.yml — database connection
views/*.view.yml — semantic layer definitions
CLAUDE.md — project instructions for Claude Code
.claude/ — agents and skills for Claude Code

airlayer init

Claude Code integration

Airlayer generates Claude Code agents and skills for working with your semantic layer:

analyst— answers data questions via semantic queries
builder— creates and edits view definitions

Skills include bootstrap, profile, and query for schema discovery, data profiling, and semantic queries.