LineageSutra AI
Column lineage from SQL text. Not a connection to a database.
Offline SQL lineage and schema-change impact CLI. npm [email protected] published 2026-11-17. The global command alias is lineagesutra. The README says it reads SQL text or a dbt manifest and never opens a database, never reads a row, and never makes a network call in scan, impact, compare, contract, verify, audit, or mcp. On 2026-11-17 the npm downloads API returned no row, so both counts are 0. GitHub showed 0 stars and 0 open issues. The name LineageSutra has not been trademark-cleared.
- Offline
- 25 skills
- Node 20+
- npm latest
- [email protected]
- GitHub version
- 0.2.0
- Licence
- Apache-2.0
- Language
- JavaScript
- Weekly downloads
- 0
- Overall downloads
- 0
- GitHub stars
- 0
Live npm weekly downloads, npm overall downloads, and the latest npm version, plus GitHub stars and the package.json version, fetched 2026-11-17 19:07 UTC. Cached 10 minutes. Weekly downloads cover 2026-11-10 to 2026-11-16. Overall downloads cover 2026-10-01 to 2026-11-17. npm does not publish an earlier total. Stars and downloads show public use. They are not a quality score.
Use it when
- 1
You have a folder of SQL, or a dbt manifest.json, and want column edges with file and line.
- 2
You want a rename, drop, or type-change rehearsal that keeps confirmed paths apart from possible ones.
- 3
You do not want the tool to open a database or read a row.
What it does, and what it does not
Read this before the guide. A limit here is a limit of the tool.
Does
- scan builds table and column edges from CREATE TABLE, CREATE VIEW, CREATE TABLE AS SELECT, and INSERT ... SELECT, with file, line, and a direct or derived kind.
- impact rehearses rename, drop, type, or modify. Confirmed means a static column path. Possible means a filter, join, group-by, or table-level dependent that is not proven at field level.
- upstream walks a column back to its origin. compare lists columns and datasets removed between two scans and who consumed them. Exit 3 is a gate you asked for.
- contract checks required columns, types, NOT NULL, and forbidden columns against DDL in the scan. The README says freshness is reported not-run and is never filled in.
- dbt imports metadata from manifest.json (nodes, sources, exposures, depends_on) and does not run dbt. The README says that import was tested on a synthetic manifest.
- audit writes Markdown, HTML, and JSON. MCP exposes read-only stdio tools scoped to one folder. The README says the MCP check used a scripted client, not MCP Inspector.
- install-skills copies 25 skills and refuses to overwrite. The README says native activation in each host is not tested. 0 runtime dependencies. Node.js 20 or newer.
Does not
- It does not connect to a database, read table rows, or see SQL that exists only at runtime, in an ORM, or only in production.
- MERGE, UPDATE, DELETE, stored procedures, dynamic SQL, Jinja inside SQL, PIVOT, UNNEST, recursive CTE self-reference, ALTER TABLE replay, and SELECT * EXCEPT stay warnings. The README says no dialect is certified. A SQLGlot adapter is planned and not built.
- A name such as email is a name hint. The README says it is not a PII classification.
- The README reports 54 of 54 exact matches and 100% precision and recall on 148 expected edges in a corpus the author labelled before running the parser. It says that corpus is not an independent benchmark. This page did not re-run it.
- SHA-256 detects an edit. The README says it is not a signature and proves nothing about who made the file. A passing report is not a certification.
- Contract checks use declared metadata, not live tables. Freshness is not measured. No telemetry. The name LineageSutra has not been trademark-cleared.
Claims above are from the project pages and are attributed as stated there. They are not independently verified here.
How a run moves
From the LineageSutra AI README, npm 0.2.0, published 2026-11-17. It reads SQL files or a dbt manifest. It does not open a database.
-
1. Scan
Text only
CREATE TABLE, CREATE VIEW, CTAS, and INSERT ... SELECT. Edges keep a file and a line. What the parser cannot read becomes a warning.
-
2. Impact
Two lists
Confirmed is a static column path. Possible is a filter, join, group-by, or table-level use that is not proven at field level.
-
3. Contract
Declared metadata
Required columns, types, NOT NULL, and forbidden columns. Freshness stays not-run. Production tables are not queried.
-
4. Seal
SHA-256
verify detects an edited report. The README says the seal is not a signature. A pass is not a certification.
How to read the result
Wording is from the project page or README and is attributed as stated there. It is not independently verified here.
Commands and notes
-
npx [email protected] audit --out ./lineage-auditself-audit report bundle. -
lineagesutra scan --root ./sql --out lineage.jsongraph from SQL text. --strict exits 3 when something was not parsed. -
lineagesutra impact --input lineage.json --table customers --column email --change renameconfirmed paths versus possible ones. -
lineagesutra upstream, compare, contract, dbt, and verifylisted in the README command table. -
lineagesutra mcp --root ./sqlread-only MCP tools. The folder is fixed at start. serve binds loopback and needs LINEAGESUTRA_TOKEN.
Guide
How to use the result. GitHub is the source repository. npm is the install.
What 0.2 reads
From the LineageSutra AI README, npm 0.2.0, published 2026-11-17. Node.js 20 or newer. 0 runtime dependencies. The command alias is lineagesutra.
- npx [email protected] scan --root ./sql builds column edges from CREATE TABLE, CREATE VIEW, CTAS, and INSERT ... SELECT, with file and line.
- impact splits a rename, drop, or type change into confirmed column paths and possible uses (filters, joins, group-bys, table-level dependents).
- dbt imports manifest.json metadata and does not run dbt. contract checks declared columns. The README says freshness is not-run and is never filled in.
- Exit 0 is ok. Exit 1 is a failed audit. Exit 2 is bad input. Exit 3 is a gate you asked for, such as --strict or --fail-on-breaking.
What a graph is not
- The README says the tool never opens a database and never reads a row. Runtime SQL, ORM SQL, and what is deployed stay invisible.
- MERGE, dynamic SQL, Jinja, procedures, and the other unsupported forms stay warnings. No dialect is certified. SQLGlot is not built.
- The README reports 54 of 54 exact matches on an author-labelled corpus of 148 edges and says that corpus is not an independent benchmark. This site did not re-run it.
- SHA-256 detects an edit and is not a signature. MCP was checked with a scripted client, not MCP Inspector. 25 skills copy into a project. IDE activation is not tested. The name LineageSutra has not been trademark-cleared. On 2026-11-17 npm had no download row, so the snapshot count is 0.
Documentation
Titles below are the HTML titles from the GitHub Pages site on 2026-11-17. A sentence is shown only when that page has its own meta description. This website did not re-check the measurements those pages describe.
- Documentation · LineageSutra AI
- Executive summary · LineageSutra AI
- Functional specification · LineageSutra AI
- Architecture · LineageSutra AI
- Technical reference · LineageSutra AI
- User guide · LineageSutra AI
- Security and privacy · LineageSutra AI
- Evidence model and rules · LineageSutra AI
- Test report · LineageSutra AI
- Operations and release · LineageSutra AI
- Legal and governance · LineageSutra AI
- Roadmap and FAQ · LineageSutra AI
- Glossary · LineageSutra AI
- Assurance and evidence · LineageSutra AI
Questions
Does LineageSutra connect to my warehouse?
No. The README says there is no API key, no network call in scan, impact, compare, contract, verify, audit, or mcp, no database connection, and no row data. It analyses SQL text or a dbt manifest.json and does not run dbt.
Does 54 of 54 mean my SQL will parse?
No. The README says that score is 54 hand-labelled cases and 148 expected edges, labelled by the author before the parser ran, and that it is a regression corpus, not an independent benchmark. Passing it does not mean the parser is correct on your SQL. audit --root lists what it could not parse. This page did not re-run the corpus.
Other projects
- VigilSutra AI 256 last week · 2,190 overall · 4 stars
- RachnaSutra AI 208 last week · 925 overall · 4 stars
- ArchSutra AI 176 last week · 883 overall · 4 stars
- PramanaSutra AI 195 last week · 750 overall · 4 stars
- EvalSutra AI 207 last week · 563 overall · 4 stars
- DocSutra AI 511 last week · 511 overall · 4 stars
- IncidentSutra AI 490 last week · 490 overall · 4 stars
- ContextSutra AI 529 last week · 529 overall · 4 stars
- MigraSutra AI 519 last week · 519 overall · 4 stars
- WattSutra AI 440 last week · 440 overall · 4 stars
- ProtocolSutra AI 530 last week · 530 overall · 4 stars
- EdgeSutra AI 337 last week · 337 overall · 4 stars
- ContribSutra AI 274 last week · 274 overall · 4 stars
- LocalSutra AI 295 last week · 295 overall · 4 stars
- DeploySutra AI 0 last week · 0 overall · 0 stars