Configuration

CLI: lineagescope scan

Option	Description
path	Directory to scan (default: .).
-d, --dialect	SQLGlot dialect (snowflake, bigquery, postgres, duckdb, …).
-f, --format	terminal (default) or json.
-e, --exclude	Comma- or semicolon-separated directory names to prune while walking (case-insensitive). Hidden directories (.*) are always skipped.
--dead-asset-whitelist	Comma-separated asset names excluded from dead-asset analysis.
--dead-asset-terminal-tags	Comma-separated substrings matching tag keys/values for intentional sinks (default behavior uses exposure, dashboard, export; pass empty string to disable).
--test-coverage-critical-deps	Downstream threshold for CRITICAL missing-test findings (default: 10).

lineagescope diff and lineagescope ci accept --path, --dialect, and --exclude where applicable.

Environment variables

Variable	Used by	Description
LINEAGESCOPE_SNAPSHOT_RETENTION_DAYS	scan, ci	Days to keep timestamped snapshot files under .lineagescope/snapshots/ (daily YYYY-MM-DD.json files are not pruned by age). Invalid values fall back to 30. Legacy: PIPESCOPE_SNAPSHOT_RETENTION_DAYS is still read if unset.
GITHUB_TOKEN	ci	If set in GitHub Actions with GITHUB_REF pointing at a PR, LineageScope may post a short score comment (optional).
GITHUB_REF, GITHUB_REPOSITORY	ci	Used together with GITHUB_TOKEN for PR comment detection.
PYTHONUTF8	CLI (Windows)	Set to 1 for reliable UTF-8 in consoles (LineageScope also tries to reconfigure stdio on Windows).

Parse warnings

If a file cannot be read or a parser raises, LineageScope skips that file and records a message in parse_warnings (JSON) or a yellow panel (terminal). The scan continues.

dbt ownership and partitions

In schema.yml (or other dbt schema files LineageScope reads):

• meta.owner on models and source tables feeds the ownership analyzer.
• meta.partition_key tags models/tables for partition-aware cost checks.
• Column data_type improves contract and typing alignment.

CODEOWNERS

Place .github/CODEOWNERS or CODEOWNERS at the repo root (or under the scan root). GitHub-style globs apply; last matching line wins.

Data contracts (ODCS)

YAML files under the scan tree that look like ODCS (e.g. dataContractSpecification, kind: DataContract, top-level schema: lists) are matched to assets by name and compared for columns/types.