API: Catalog

Business glossary, asset certification, ownership, and lineage. All endpoints require authentication unless noted. 38 endpoints, 4 of them app only: an API key gets 403 there.

GET /api/audit-logs

List audit logs

Query parameters

Name	Type	Required	Description
userId	string	no
action	string	no
resourceType	string	no
from	string	no
to	string	no
limit	integer	no
offset	integer	no

Responses: 200 · 400 · 401

GET /api/audit-logs/export

Export audit logs (CSV download, or seq-cursored NDJSON for SIEM collectors)

Part of the Enterprise plan (compliance auditing): below it 403 PLAN_TIER_REQUIRED, unless the workspace keeps the export until the renewal of the period running when it moved. The NDJSON feed also carries the workspace’s rows kept in the 7-year archive (each line’s kept: archive or live). Reading the log stays on every plan.

Query parameters

Name	Type	Required	Description
userId	string	no
action	string	no
resourceType	string	no
from	string	no
to	string	no
limit	integer	no
format	"csv" | "ndjson"	no
afterSeq	integer	no

Responses: 200

GET /api/audit-logs/queries

List query executions: who ran which query on which source, how long, how many rows, and any error

Query parameters

Name	Type	Required	Description
userId	string	no
sourceId	string	no
origin	string	no
errorsOnly	boolean	no
from	string	no
to	string	no
limit	integer	no
offset	integer	no

Responses: 200 · 400 · 401

GET /api/audit-logs/queries/export

Export query executions as NDJSON for SIEM collectors, oldest first, resumed with the x-next-cursor header

Part of the Enterprise plan (compliance auditing): below it 403 PLAN_TIER_REQUIRED, unless the workspace keeps the export until the renewal of the period running when it moved. Listing query executions stays on every plan.

Query parameters

Name	Type	Required	Description
limit	integer	no
after	string	no

Responses: 200

GET /api/audit-logs/verify

Verify the audit log hash chain

Query parameters

Name	Type	Required	Description
fromSeq	integer	no
limit	integer	no

Responses: 200

GET /api/catalog/assets

List governed assets with owner/domain facets

The dashboards, models, data sources and metrics the reader may open, each with its owner, domain, certification and where it opens. A request with none of q, sort, order or offset answers as it always has: { items, facets }, up to 500 by type then name, facets over the type asked. A request with any of them answers a page: { items, total, facets }, 50 by default, sorted by sort (name) in order (asc), names whatever their case, no owner or never changed last either way, ties by name, then type, then id; total counts the matches on every page, the facets describe the whole workspace whatever the filters, and facets.hidden counts the assets of types the reader may not open, never naming them.

Query parameters

Name	Type	Required	Description
type	"dashboard" | "model" | "source" | "metric"	no
domain	string	no
ownerId	string	no
unowned	boolean	no
undocumented	boolean	no	Only what Describe with AI can document: dashboards and metrics with no description, and models with a visible table or field that has none.
q	string	no	Text the asset’s name holds, whatever its case (a % or _ is that character); close spellings too where the database can.
sort	"name" | "type" | "updated" | "owner"	no	A page’s order: name unless given.
order	"asc" | "desc"	no	asc unless given. No owner, or never changed, comes last either way.
offset	integer	no
limit	integer	no	500 on today’s call, 50 on a page.

Schema: CatalogAssetsPage (the response to a page request, one with q, sort, order or offset)

Responses: 200 · 400 · 401 · 403

PATCH /api/catalog/assets/{type}/{id}

Assign owner/domain to a governed asset

Sets or clears the asset’s owner, its business domain, or both (null clears one; a body with neither is refused). The owner must be an active member of the workspace (400, details.reason owner_not_a_member). A domain is trimmed, each run of spaces one space, and written in the spelling the workspace already uses whatever its case. A model’s owner also reviews the AI’s notes about the model, so only someone who can edit models sets or clears it (403, details.reason caller_cannot_edit_models), and only to someone who can edit models (400, details.reason owner_cannot_edit_models). Setting an owner or a domain does not change the asset’s updated_at.

Path parameters

Name	Type	Required	Description
type	"dashboard" | "model" | "source" | "metric"	yes
id	string	yes

Request body (JSON)

Field	Type	Required	Description
ownerId	string | null	no	The new owner, a member of the workspace. Null clears it; absent leaves it.
domain	string | null	no	The new domain, in the spelling the workspace already uses. Null or blank clears it; absent leaves it.

Schema: CatalogAssetPatch (the request body) · CatalogAssetPatchAnswer (the response)

Responses: 200 · 400 · 401 · 403 · 404

GET /api/catalog/assets/{type}/{id}/notices

Data notices: an asset’s own and those it inherits

For a dashboard, model, data source or metric the caller may open: its own notices, then those on what it is built on, every step of the way, as the lineage graph reads it. What the caller may not open is never named: such a notice’s asset has no id, name or link, and no setter. can_manage says whether the caller may set, edit and clear its own; setting_needs_upgrade, that setting and editing need Business. Every plan. The first 50, with total. 404 when the caller can’t open the asset. An API key needs dashboard:read, and source:read for a data source, as the Catalog opens one.

Path parameters

Name	Type	Required	Description
type	"dashboard" | "model" | "source" | "metric"	yes
id	string	yes

Responses: 200 · 400 · 401 · 403 · 404

POST /api/catalog/assets/{type}/{id}/notices

Set a data notice on an asset

Sets a notice on a dashboard, model, data source or metric. Its owner, someone who may edit it (the type’s own update; for a dashboard an editor grant on it too) or an admin (403, details.reason cannot_manage otherwise), on an asset they may open (404 otherwise). Business (403 PLAN_TIER_REQUIRED on Team). One notice of each kind per asset: 409 while one of that kind is in force (edit it instead); one that has ended gives its place to this one. 400 for words past the limit or holding hidden text, a custom note without words, or an end that has passed. Its words show to everyone who reads what is built on it; its asset’s name only to those who may open it. An API key needs dashboard:write, and its owner’s role decides.

Path parameters

Name	Type	Required	Description
type	"dashboard" | "model" | "source" | "metric"	yes
id	string	yes

Request body (JSON)

Field	Type	Required	Description
kind	"deprecated" | "maintenance" | "stale" | "sensitive" | "custom"	yes	deprecated (being retired: use something else), maintenance (being worked on: its numbers may change), stale (its data is out of date), sensitive (it holds sensitive data: take care sharing it), or custom (a note in the setter’s own words).
message	string | null	no	Its words, plain text: white space is folded to single spaces and the result kept to 500 characters. Required for a custom note; null or blank says no more than its kind.
expiresAt	string | null	no	When it stops showing on its own, an instant still to come. Null or absent: until someone clears it.

Responses: 201 · 400 · 401 · 403 · 404 · 409

PATCH /api/catalog/assets/{type}/{id}/notices/{noticeId}

Edit a data notice

Changes a notice’s words, its end, or both: null clears one, absent leaves it. Its kind stays. Its owner, someone who may edit it (the type’s own update; for a dashboard an editor grant on it too) or an admin (403, details.reason cannot_manage otherwise), on an asset they may open (404 otherwise). Business (403 PLAN_TIER_REQUIRED on Team). 404 for a notice that has ended, which shows nowhere. 400 for an empty change, words past the limit or holding hidden text, a custom note without words, or an end that has passed. An API key needs dashboard:write, and its owner’s role decides.

Path parameters

Name	Type	Required	Description
type	"dashboard" | "model" | "source" | "metric"	yes
id	string	yes
noticeId	string	yes

Request body (JSON)

Field	Type	Required	Description
message	string | null	no	Its words, plain text: white space is folded to single spaces and the result kept to 500 characters. Required for a custom note; null or blank says no more than its kind.
expiresAt	string | null	no	When it stops showing on its own, an instant still to come. Null or absent: until someone clears it.

Responses: 200 · 400 · 401 · 403 · 404

DELETE /api/catalog/assets/{type}/{id}/notices/{noticeId}

Clear a data notice

Clears a notice: it shows nowhere from now on, and the audit log keeps what it said. Its owner, someone who may edit it (the type’s own update; for a dashboard an editor grant on it too) or an admin (403, details.reason cannot_manage otherwise), on an asset they may open (404 otherwise). Every plan, so a notice left after a downgrade can always come down. An API key needs dashboard:write, and its owner’s role decides.

Path parameters

Name	Type	Required	Description
type	"dashboard" | "model" | "source" | "metric"	yes
id	string	yes
noticeId	string	yes

Responses: 204 — Cleared · 400 · 401 · 403 · 404

GET /api/catalog/assets/{type}/{id}/summary

About this data: one asset’s trust card

For a dashboard, model, data source or metric the caller may open: its owner, domain, certification (who, when, and a metric’s note and version), description, when it last changed, how fresh the data it reads is, the latest outcome of the quality checks on what it reads, a dashboard’s views in the last 30 days, the glossary terms linked to it, and how many assets feed it and read it, every step of the way: what the caller may open counted by type, the rest counted and never named. explain says whether Explain this asset would run for the caller; asking runs no AI and spends nothing. Every plan; lineage and Explain this asset need Business (withheld.lineage is plan), and lineage the Catalog’s view (permission). 404 when the caller can’t open the asset or the workspace has no such asset. An API key needs dashboard:read.

Path parameters

Name	Type	Required	Description
type	"dashboard" | "model" | "source" | "metric"	yes
id	string	yes

Responses: 200 · 400 · 401 · 403 · 404

GET /api/catalog/descriptions/drafts

Describe with AI: the drafts waiting for your review, and whether it would run for you now

App only. An API key gets 403: AI features are available in the DataSquares app; API keys can’t use them.

The AI-drafted descriptions of the dashboards and metrics the caller may edit: state=pending (the default) lists the ones waiting for review, by type then name; state=all adds the decided ones (published: what an approval wrote), newest first. type and id narrow it to one asset. A pending draft whose asset got a description another way meanwhile is superseded. describeWithAi says whether drafting would run for the caller now. No AI is called.

Query parameters

Name	Type	Required	Description
state	"pending" | "all"	no
type	"dashboard" | "metric"	no
id	string	no

Responses: 200 · 400 · 401 · 403 · 404

POST /api/catalog/descriptions/drafts

Describe with AI: draft the missing descriptions of dashboards and metrics (AI credits)

App only. An API key gets 403: AI features are available in the DataSquares app; API keys can’t use them.

Drafts a description for each dashboard and metric named (at most 10) that has none, one quick AI call each, charged in AI credits (Description drafts). The AI is sent what the asset is, from its lineage chain: its name, a dashboard’s card titles, a metric’s formula (quoted text left out), grain and time field, and the models, fields, measures, metrics and filters it reads, with the descriptions they have; never a value, and a labelled field by its name alone. Nothing is published: each draft waits for a person’s review. Each asset’s outcome: drafted (the draft waits for review), pending (a draft already waited: no AI call), described (it has a description: no AI call), empty (a dashboard with no cards: no AI call), dropped (the checks left the draft out, its text not kept; reason: too_long, screened (personal, secret or instruction-like text), invented_number (a number the asset does not carry), no_answer), failed (the AI call failed: nothing stored) or not_started (the request’s time ran out first: ask again). Drafting and reviewing take the asset’s own edit right (403 details.reason cannot_edit, naming them); several assets in one request take the Catalog’s update right too (403 details.reason several_need_catalog); an asset the caller may not open is not found (404). One request at a time per workspace: another meanwhile is refused with 409 DRAFTS_RUNNING before any AI call, details.running saying whether you started the run, when, and for how many assets; 429 DRAFTS_BUSY (Retry-After 60 seconds) when the server drafts as many runs as it holds at once. 503 AI_NOT_CONFIGURED on a server without AI.

Request body (JSON)

Field	Type	Required	Description
assets	array of object	yes

Responses: 200 · 400 · 401 · 403 · 404 · 409 · 429 · 503

PUT /api/catalog/descriptions/drafts

Describe with AI: approve or reject drafted descriptions

App only. An API key gets 403: AI features are available in the DataSquares app; API keys can’t use them.

approve writes the draft, or text in its place, as the dashboard’s or metric’s description, only where it still has none (otherwise the draft is superseded); reject keeps it out. Every draft must be one of the workspace’s drafts waiting for review (400 otherwise, naming them), of an asset the caller may edit (403 otherwise, details.reason cannot_edit), and nothing is changed then.

Request body (JSON)

Field	Type	Required	Description
decisions	array of object	yes

Responses: 200 · 400 · 401 · 403 · 404

GET /api/catalog/notices/for/{type}/{id}

Data notices for a viewed dashboard, report, presentation or story

The notices in force on what a dashboard, report, presentation or story the caller may open is built on, every step of the way through the lineage graph (a notice on a data source reaches the dashboards whose models read it, and so on), and a dashboard’s own; for the short line its page shows. Read as the caller reads the item: a presentation’s speaker notes and working copy for its makers, its published version for everyone else; a story’s sections they may see. What the caller may not open is never named: such a notice’s asset has no id, name or link, no setter or editor, and an id of its own in each answer. Every plan. 404 when the caller can’t open the item. An API key needs the item’s own read scope (dashboard:read, reports:read, presentations:read, stories:read), and names an asset only with dashboard:read.

Path parameters

Name	Type	Required	Description
type	"dashboard" | "report" | "presentation" | "story"	yes
id	string	yes

Responses: 200 · 400 · 401 · 403 · 404

PUT /api/certification/{assetType}/{id}

Certify or un-certify a governed asset (workspace admins)

Workspace admins. A metric may also be certified by someone whose role holds the Certify metrics permission; its certification records the version it is of (certified_version) and an optional note, both cleared by its next version.

Path parameters

Name	Type	Required	Description
assetType	"dashboard" | "model" | "source" | "metric"	yes
id	string	yes

Request body (JSON)

Field	Type	Required	Description
certified	boolean	yes
note	string | null	no	A metric’s certification only: why it is certified, such as “for board reporting”. Kept with it until a new version clears it.

Responses: 200 · 400 · 401 · 404

GET /api/classification/proposals

A model’s open sensitive-column proposals

Every open proposal of a field still unlabelled and still proposable (visible, not calculated, of a visible table and a classified source), by table and field order: the detector its sampled values matched, the label proposed, how many values were sampled and how many matched, and whether the column’s name agreed. No value is kept or returned. With where the model’s checks stand: when a table was last read, how many tables the last check could not read, how many columns are not checked because their names cannot be written safely in a query, and how many unlabelled columns without a proposal are not checked because a column of the same name carries a label elsewhere on the source (that label does not mask them), with the first 200 of them by table and field.

Query parameters

Name	Type	Required	Description
modelId	string	yes	The model whose proposals to list.

Responses: 200 · 400 · 401 · 403 · 404

POST /api/classification/proposals/approve

Approve sensitive-column proposals: add their labels (model editors, up to 200 at once)

Checked whole first: 404 when a proposal is not this model’s, 409 when one was decided already or is no longer proposed (its column or table was hidden, deleted or changed since, or its live database is no longer checked), and nothing is changed. Each approval then adds the proposed label to its field in one transaction with the decision; a field that carries a label by then keeps it (a model editor may add a label, never change one) and its proposal is closed as superseded. One that can no longer be applied by then (decided, withdrawn, or no longer proposed meanwhile) is left as it is and counted in notApplied. Should a write fail, the approvals before it stay and the rest stay open.

Request body (JSON)

Field	Type	Required	Description
modelId	string	yes	The model the proposals are of.
ids	array of string	yes	The proposals to decide: 1 to 200, each once.

Responses: 200 · 400 · 401 · 403 · 404 · 409

POST /api/classification/proposals/reject

Reject sensitive-column proposals (model editors, up to 200 at once)

All or nothing, as approving. A rejected field is not proposed again for 90 days; its label is not changed.

Request body (JSON)

Field	Type	Required	Description
modelId	string	yes	The model the proposals are of.
ids	array of string	yes	The proposals to decide: 1 to 200, each once.

Responses: 200 · 400 · 401 · 403 · 404 · 409

GET /api/classification/settings

Whether this workspace’s live databases are classified for sensitive columns

Sensitive columns are found by rules over sampled values (never an AI). Staged sources and the Warehouse are always classified; live databases only once a company admin turns it on.

Responses: 200 · 400 · 401 · 403 · 404

PUT /api/classification/settings

Turn classification of live databases on or off (company admins)

On, the sampler also reads this workspace’s live databases: one bounded query a table, run by the system on the database’s own server, at most 200 values a column, matched and dropped; only counts are kept.

Request body (JSON)

Field	Type	Required	Description
liveSources	boolean	yes

Responses: 200 · 400 · 401 · 403 · 404

GET /api/company-audit

The company audit view: every workspace’s actions and the company-level ones (Enterprise)

Every audit row of your company, newest first: each workspace’s, and the actions that belong to no workspace (people, the company profile, its security policy, single sign-on, menus, billing, sign-ins). Two years, as every audit log keeps; the export reads the 7-year archive too. Company admins only, on the Enterprise plan.

Query parameters

Name	Type	Required	Description
scope	"all" | "company"	no	company: only actions with no workspace (people, settings, sign-ins).
workspaceId	string	no	Only this workspace’s rows (one of your company’s).
userId	string	no
action	string	no
resourceType	string	no
from	string	no
to	string	no
limit	integer	no
offset	integer	no

Responses: 200 · 400 · 401

GET /api/company-audit/export

Export the company’s audit trail, the 7-year archive included: CSV, or NDJSON for SIEM collectors (Enterprise)

For a signed-in company admin, or a collector polling with an API key (X-API-Key) whose owner is a company admin and whose scopes let it read the audit log (admin_audit:view, or read).

Query parameters

Name	Type	Required	Description
scope	"all" | "company"	no	company: only actions with no workspace (people, settings, sign-ins).
workspaceId	string	no	Only this workspace’s rows (one of your company’s).
userId	string	no
action	string	no
resourceType	string	no
from	string	no
to	string	no
format	"csv" | "ndjson"	no
afterSeq	integer	no	NDJSON: the x-last-seq of the page before; absent = from the oldest kept row.
limit	integer	no	CSV: at most this many rows (more is refused). NDJSON: one page (at most 10,000).

Responses: 200

GET /api/glossary

List glossary terms

Every live term with its links, in name order (the default, as before). view=lookup: each term’s name, synonyms and a definition cut to 280 characters, no links (for field tooltips). view=page: \{ items, total, offset, limit }, found by q in a name, a synonym or a definition. A link to a metric, a model or a field is listed only to a reader who may open metrics, or models. Every view sends an ETag; send it back in If-None-Match to get 304 when nothing changed.

Query parameters

Name	Type	Required	Description
view	"full" | "lookup" | "page"	no
q	string	no	view=page: words in the name, a synonym or the definition.
offset	integer	no
limit	integer	no

Schema: GlossaryTerm (the response without view: a list of them) · GlossaryPage (the response to view=page) · GlossaryLookupItem (the response to view=lookup: a list of them)

Responses: 200 · 304 — Not modified · 400 · 401 · 403

POST /api/glossary

Create a glossary term

Published at once. 409 CONFLICT (details.reason term_taken, with the term) for a name the workspace has; 400 VALIDATION_ERROR for a blank name or definition (blank), characters that can’t be seen (hidden_text), a word another term already has (synonym_taken, with the word and the term) or a steward who is not a member (not_a_member).

Request body (JSON)

Field	Type	Required	Description
term	string	yes
definition	string	yes
synonyms	array of string	no
stewardId	string | null	no	The person to ask about the term (a member of the workspace); null clears it.

Schema: GlossaryTermCreate (the request body) · GlossaryTerm (the response)

Responses: 201 · 400 · 401 · 403 · 409

PUT /api/glossary/{id}

Update a glossary term

Changes the fields sent and answers the term with the links the reader may open; an edit that changes nothing writes nothing. Send expectedUpdatedAt (the term’s updated_at as read) to be refused (409 CONFLICT, details.reason changed) when someone changed it since. 409 term_taken for a name the workspace has; 400 as for create, for the names the edit adds only; 404 when there is no such term.

Path parameters

Name	Type	Required	Description
id	string	yes

Request body (JSON)

Field	Type	Required	Description
term	string	no
definition	string	no
synonyms	array of string	no
stewardId	string | null	no	The person to ask about the term (a member of the workspace); null clears it.
expectedUpdatedAt	string	no	The term’s updated_at when the editor opened it: a 409 (details.reason changed) when it changed since.

Schema: GlossaryTermUpdate (the request body) · GlossaryTerm (the response)

Responses: 200 · 400 · 401 · 403 · 404 · 409

DELETE /api/glossary/{id}

Delete a glossary term

On every plan, for a role that may delete glossary terms. 404 when there is no such live term.

Path parameters

Name	Type	Required	Description
id	string	yes

Responses: 204 — Deleted · 400 · 401 · 403 · 404

GET /api/glossary/{id}/history

A glossary term’s history

Its changes, newest first, from the audit log (kept two years): created, updated, deleted, link added, link removed, with who and what each changed; a link’s change only to a reader who may open its kind of asset. A deleted term’s history is readable too. On every plan.

Path parameters

Name	Type	Required	Description
id	string	yes

Query parameters

Name	Type	Required	Description
limit	integer	no

Schema: GlossaryHistory (the response)

Responses: 200 · 400 · 401 · 403 · 404

POST /api/glossary/{id}/links

Link a glossary term to an asset

The asset must be a live metric, model or field of the workspace, of a kind the caller may open (400 VALIDATION_ERROR otherwise, details.reason asset_not_found with its asset_type, malformed, blank or not_a_field_link). Answers the link with its term and its asset’s name; linking what is already linked answers that link.

Path parameters

Name	Type	Required	Description
id	string	yes

Request body (JSON)

Field	Type	Required	Description
assetType	"metric" | "model_field" | "model"	yes
assetRef	string	no	A metric’s or a model’s id, or a field’s ‘<model id>:<field name>’.
fieldId	string	no	model_field only: the field itself (instead of assetRef).

Schema: GlossaryLinkCreate (the request body) · GlossaryLinkCreated (the response)

Responses: 201 · 400 · 401 · 403 · 404 · 409

DELETE /api/glossary/{id}/links/{linkId}

Unlink a glossary term from an asset

404 when the term has no such link, or none of a kind of asset the caller may open.

Path parameters

Name	Type	Required	Description
id	string	yes
linkId	string	yes

Responses: 204 — Unlinked · 400 · 401 · 403 · 404

GET /api/glossary/suggestions

Suggest terms to define: the words your names use that the glossary does not define yet

The words and two-word phrases the names of the dashboards, metrics and models (their tables, visible fields and measures) the caller may open use most, that no term names by its name or a synonym (case, plurals and separators aside), most used first. uses: how many names use it; examples: a few of them. A plain count (method count): no AI is called and no AI credits are used. Words that mean nothing alone (of, id, date, total, …) are left out; a phrase used once is left out. Only a person writes a definition.

Query parameters

Name	Type	Required	Description
limit	integer	no

Responses: 200 · 400 · 401 · 403

GET /api/lineage/graph

Asset-level lineage graph (source → pipeline → model or blend → query or metric → dashboard or report → share, and the decks and stories that bind a metric, a model or a saved query)

Names only what the caller may open, by the same permissions as the pages that open each asset (sources, pipelines, models and blends, saved queries, metrics, dashboards and reports by role or grant, decks and stories by their own readers); the rest is counted, one node per neighbour, with hidden holding how many (“2 sources you can’t open”), and it opens nothing. With asset, the view is that asset’s lineage: what feeds it (direction=up), what reads it (down) or both (the default), depth hops from it (1 to 10, 2 by default) or all; without asset both are ignored and the graph is the whole workspace. At most 5000 assets are drawn (a scoped view keeps the nearest first); truncated is true when the depth or that limit left some out. An asset the caller may not open, or one that does not exist, is an empty graph. Requires the Catalog view permission; an API key also needs dashboard:read.

Query parameters

Name	Type	Required	Description
asset	string	no
direction	"up" | "down" | "both"	no
depth	string	no

Schema: LineageGraph (the response)

Responses: 200 · 400 · 401 · 403 · 429

GET /api/lineage/graph/chain

The lineage chain of a dashboard, a card, a metric or a model

Built by rules from the workspace’s references, in a fixed order: source, pipeline, Warehouse dataset, model, the fields, measures and metric formulas read, the cards, the dashboard, then its filters. A model’s chain is what feeds it, up to the model itself. Only what the caller may open is named; the rest is counted in hidden. A model a row-level security policy applies on is marked policy (never its rule); a labelled field is marked labelled; no value is included. Column lineage through pipelines is not tracked, and a chain with a pipeline says so in notes. explainWithAi says whether Explain with AI would run for the caller now, and if not why. 404 when the caller can’t open the dashboard (or the card’s), or the model (the Data models view).

Query parameters

Name	Type	Required	Description
kind	"dashboard" | "card" | "metric" | "model"	yes
id	string	yes

Schema: LineageChainAnswer (the response)

Responses: 200 · 404

GET /api/lineage/graph/count

How many assets the lineage graph names, of each type, without the graph

The view GET /api/lineage/graph would draw for the caller, counted: total the assets it names (only what the caller may open, by the same permissions), byType how many of each type, a type with none left out. asset, direction and depth scope it as they scope the graph. Never cut short: the graph draws at most 5000 assets, the count counts them all. Requires the Catalog view permission; an API key also needs dashboard:read.

Query parameters

Name	Type	Required	Description
asset	string	no
direction	"up" | "down" | "both"	no
depth	string	no

Responses: 200 · 400 · 401 · 403 · 429

POST /api/lineage/graph/explain

Explain the lineage of a dashboard, a card, a metric or a model in one paragraph (AI)

App only. An API key gets 403: AI features are available in the DataSquares app; API keys can’t use them.

Uses AI credits. One AI call writes a paragraph from the lineage chain alone (the chain GET /chain returns), charged to the caller for the AI work it used. The paragraph is checked: every name in it must come from the chain and every number must be one of its counts. outcome is written when it passes, dropped when it does not (paragraph is null and the chain is the answer), and unavailable when the AI returned nothing usable. Nothing is stored. 400 when there is no lineage to explain; 503 AI_NOT_CONFIGURED when AI isn’t set up on this server.

Request body (JSON)

Field	Type	Required	Description
kind	"dashboard" | "card" | "metric" | "model"	yes
id	string	yes

Schema: LineageExplanation (the response)

Responses: 200 · 400 · 403 · 404 · 429 · 503

PUT /api/ownership/{assetType}/{id}

Set/clear an asset’s owner and business domain

The same assignment, body and answer as PATCH /api/catalog/assets/:type/:id: null clears the owner or the domain. The owner must be an active member of the workspace (400, details.reason owner_not_a_member); a model’s owner is set only by someone who can edit models (403, details.reason caller_cannot_edit_models), to someone who can edit models (400, details.reason owner_cannot_edit_models).

Path parameters

Name	Type	Required	Description
assetType	"dashboard" | "model" | "source" | "metric"	yes
id	string	yes

Request body (JSON)

Field	Type	Required	Description
ownerId	string | null	no	The new owner, a member of the workspace. Null clears it; absent leaves it.
domain	string | null	no	The new domain, in the spelling the workspace already uses. Null or blank clears it; absent leaves it.

Schema: CatalogAssetPatch (the request body) · CatalogAssetPatchAnswer (the response)

Responses: 200 · 400 · 401 · 403 · 404

GET /api/ownership/domains

Business domains in use (filter facets)

Responses: 200 · 400 · 401 · 403 · 404