What problem does it solve? Writing raw SQL against heterogeneous databases forces developers to juggle different drivers, parameter styles, and connection lifecycles. This Skill provides guidance for SQLSpec, a type-safe SQL query mapper (not an ORM) that unifies 15+ database adapters behind one interface with a sqlglot-powered AST pipeline for validation and dialect conversion. ## Core Features & Use Cases - Unified Adapter Layer: Configure asyncpg, psycopg, SQLite, DuckDB, Oracle, BigQuery, Spanner, and more with consistent driver methods like select_one(), select_many(), and select_to_arrow(). - Query Builder & SQL Files: Build dynamic queries fluently with sql.select() or load static SQL from files via SQLFileLoader with caching. - Framework Integrations: First-party plugins for Litestar, FastAPI, Flask, and Starlette with dependency injection, commit-mode middleware, and filter providers. - Use Case: Build a FastAPI endpoint that paginates orders from PostgreSQL using provide_session() and provide_filters(), with automatic commit on 2xx responses and typed Pydantic results. ## Quick Start Ask the AI to set up an asyncpg adapter with a typed model and run a parameterized paginated query using SQLSpec.