atlas cloud repo lingraph prints a repository’s column-level lineage to stdout, as a compact node/edge graph for rendering and traversal, or as OpenLineage RunEvents for ingestion into catalogs like Marquez and DataHub.

The column-level lineage Atlas Cloud renders in the browser is now available programmatically. The new atlas cloud repo lingraph command prints a repository's lineage graph to stdout, so you can render it, traverse it in a script, or feed it into a lineage catalog.

The machine-readable graph is also a ready source of context for AI coding agents. An agent asked to rename or drop a column can load the column-level dependency graph, trace every downstream view and column that reads from it, and either rewrite them or flag the change as breaking, without a human mapping the dependencies by hand.

A node/edge graph by default

The default format is a compact graph of nodes (tables, views, external datasets, and intermediate datasets such as CTEs) and edges. Column-level edges carry the source and target columns and the SQL projection that connects them:

atlas cloud repo lingraph --slug my-repo

$ atlas cloud repo lingraph --slug my-repo

{
  "nodes": [
    { "id": "schema/main/table/users",                "type": "table",   "name": "users",      "schema": "main" },
    { "id": "schema/main/view/user_names",            "type": "view",    "name": "user_names", "schema": "main" },
    { "id": "schema/main/dataset/user_names/u_cte@5", "type": "dataset", "name": "u_cte@5",    "schema": "main" }
  ],
  "edges": [
    { "from": "schema/main/table/users", "to": "schema/main/dataset/user_names/u_cte@5", "fromColumn": "name", "toColumn": "user_name", "expr": "u.name AS user_name" },
    { "from": "schema/main/dataset/user_names/u_cte@5", "to": "schema/main/view/user_names", "fromColumn": "user_name", "toColumn": "user_name", "expr": "u_cte.user_name AS user_name" }
  ]
}

OpenLineage for your catalog

Pass --open-lineage and Atlas emits the same lineage as a JSON array of OpenLineage RunEvents, one per view or derived dataset, ready to ingest into Marquez, DataHub, or any OpenLineage-compatible catalog. Each event carries the view's SQL, a jobType facet (VIEW or MATERIALIZED_VIEW), and a columnLineage facet that distinguishes pass-through columns (DIRECT/IDENTITY) from transformed ones (INDIRECT/TRANSFORMATION):

atlas cloud repo lingraph --slug my-repo --open-lineage

$ atlas cloud repo lingraph --slug my-repo --open-lineage

[
  {
    "eventType": "COMPLETE",
    "job": {
      "name": "active_users",
      "namespace": "atlas://my-repo/main",
      "facets": {
        "jobType": { "integration": "atlas", "jobType": "VIEW", "processingType": "BATCH" },
        "sql": { "query": "SELECT id, name FROM users" }
      }
    },
    "inputs": [ { "name": "users", "namespace": "atlas://my-repo/main" } ],
    "outputs": [
      {
        "name": "active_users",
        "namespace": "atlas://my-repo/main",
        "facets": {
          "columnLineage": {
            "fields": {
              "id":   { "inputFields": [ { "name": "users", "field": "id",   "transformations": [ { "type": "DIRECT", "subtype": "IDENTITY" } ] } ] },
              "name": { "inputFields": [ { "name": "users", "field": "name", "transformations": [ { "type": "DIRECT", "subtype": "IDENTITY" } ] } ] }
            }
          }
        }
      }
    ],
    "producer": "https://atlasgo.io",
    "run": { "runId": "82133796-1259-5d46-bb5f-92a056fa5413" }
  }
]

Managed objects use the atlas://<repo>/<schema> namespace and external sources use external://, and run.runId is deterministic, so diffs across runs are meaningful.

See the Column-Level Data Lineage docs for the full field reference and more examples.

Programmatic Access to Column-Level Lineage

A node/edge graph by default

OpenLineage for your catalog