stack.yaml

# The declarative data stack — the whole pipeline described in one file.
# Precedence: environment variable > stack.yaml > built-in default.
# (e.g. DE_ENGINE, DE_DB_PATH, DE_WINDOW_DAYS, DE_BSKY_MAX_PAGES override these.)

engine: datafusion              # duckdb | datafusion | snowflake  <- swap the engine here
db_path: de_ecosystem.duckdb

sources:                        # nouns — dlt ingests these (outside the grammar)
  bluesky:
    max_pages: 40
  github:
    slice: "data/raw/gharchive/*.json.gz"

momentum:                       # the digest metric
  window_days: 90               # 90 = laptop pulse; 365 = warehouse year-in-review
  tools: [dbt-core, dagster, dlt, ibis-framework, xorq, apache-airflow, polars,
          duckdb, great-expectations, pyiceberg, prefect, mage-ai, sqlmesh,
          soda-core, pydantic, sqlglot]