Data API

Upload files into a project, run its saved workflows and download the results, from a system of your own. Every call carries an integration key a person created under Settings, as Authorization: Bearer <key>, and acts as that person within the projects and access the key was given. File bytes never travel in these requests: an upload answers signed links to send its parts to, and an export answers a signed link to download from.

This is version 2.0.0. Keep data current from your own systems shows how the requests fit together and how to create a key. The same description is available in OpenAPI form to load into an integration tool.

Requests

List projects

GET /projects

List the projects this key reaches, to find the id of the one a script works in.

Key: Read only or Read and write.

Input	Sent in	Type	Required	Meaning
include_counts	query	boolean	no	Default: False.

Answer: 200, list of ProjectListItem.

Refused with: 401, 404, 422.

List data sources

GET /data-sources

List a project's data sources, to find the one a refreshed file goes into, or one a run wrote.

Key: Read only or Read and write.

Input	Sent in	Type	Required	Meaning
kind	query	tabular or document	no	Keep the data sources whose latest file is of this kind. Omit for all.
project_id	query	string (uuid)	yes	The project this call acts in.

Answer: 200, DataSourceListResponse.

Refused with: 401, 404, 422.

Get data source

GET /data-sources/{id}

Read one data source: its name, how many files it holds and its latest file.

Key: Read only or Read and write.

Input	Sent in	Type	Required	Meaning
id	path	string (uuid)	yes
project_id	query	string (uuid)	yes	The project this call acts in.

Answer: 200, DataSourceResponse.

Refused with: 401, 404, 422.

Upload a file

POST /data-sources/upload

Start uploading a file under a name no other file in the project has, into the data source data_source_id names (it becomes the latest) or starting one named data_source_name. The answer lists the parts to PUT to their signed links; the file's bytes never travel in this request.

Key: Read and write.

Input	Sent in	Type	Required	Meaning
name	body	string	yes	The file's name: .csv, .xlsx, .md or .txt (save .xls as .xlsx), unused by the project's files.
project_id	body	string (uuid)	yes	The project this call acts in.
sha256	body	string	yes	Lowercase hex SHA-256 over the exact whole file. Required independently of storage multipart ETags; ingestion rejects the session if the stored bytes differ.
size_bytes	body	integer	yes	The exact file size in bytes. It reserves account capacity and determines the fixed multipart slices before bytes travel.
data_source_id	body	string (uuid)	no	Joins this data source as its latest file.
data_source_name	body	string	no	Or starts a data source of this name.

Answer: 200, DataSourceUploadSessionInProgress.

Refused with: 401, 403, 404, 422.

Get upload session

GET /upload-sessions/{id}

Read where an upload stands. ready carries the new file's id and its data source's id; rejected carries what was wrong with the file.

Key: Read only or Read and write.

Input	Sent in	Type	Required	Meaning
id	path	string (uuid)	yes
project_id	query	string (uuid)	yes	The project this call acts in.

Answer: 200, DataSourceUploadSession.

Refused with: 401, 404, 422.

Resume upload

POST /upload-sessions/{id}/resume

Ask again for the signed links of the parts an upload still needs, after the earlier links expired.

Key: Read and write.

Input	Sent in	Type	Required	Meaning
id	path	string (uuid)	yes
project_id	body	string (uuid)	yes	The project this call acts in.

Answer: 200, DataSourceUploadSession.

Refused with: 401, 403, 404, 422.

Cancel data source upload

POST /upload-sessions/{id}/cancel

Abandon an upload that has not finished; nothing is published.

Key: Read and write.

Input	Sent in	Type	Required	Meaning
id	path	string (uuid)	yes
project_id	body	string (uuid)	yes	The project this call acts in.

Answer: 200, DeleteConfirmation.

Refused with: 401, 403, 404, 422.

Complete upload

POST /upload-sessions/{id}/complete

Finish an upload once every part has been sent. Checking the file takes a moment: poll the upload until it is ready or rejected.

Key: Read and write.

Input	Sent in	Type	Required	Meaning
id	path	string (uuid)	yes
project_id	body	string (uuid)	yes	The project this call acts in.

Answer: 200, DataSourceUploadSession.

Refused with: 401, 403, 404, 422.

List the files of a data source

GET /data-sources/{id}/files

List a data source's files, newest first, to see which file is its latest.

Key: Read only or Read and write.

Input	Sent in	Type	Required	Meaning
id	path	string (uuid)	yes
before_version	query	integer	no	The last number the page before answered.
limit	query	integer	no	Default: 20.
project_id	query	string (uuid)	yes

Answer: 200, list of FileResponse.

Refused with: 401, 404, 422.

List workflows

GET /workflows

List a project's workflows, to find the id of the one to run.

Key: Read only or Read and write.

Input	Sent in	Type	Required	Meaning
project_id	query	string (uuid)	yes	The project this call acts in.

Answer: 200, list of WorkflowListResponse.

Refused with: 401, 404, 422.

Run workflow

POST /workflows/{id}/run

Start a run of a saved workflow. The answer carries the execution id to follow it by.

Key: Read and write.

Input	Sent in	Type	Required	Meaning
id	path	string (uuid)	yes	UUID of the deployed workflow.
project_id	body	string (uuid)	yes	The project this call acts in.

Answer: 201, WorkflowExecutionResponse.

Refused with: 401, 403, 404, 422.

Get execution

GET /executions/{id}

Read a run's status. Poll until it has finished; a finished run lists the scenarios it published.

Key: Read only or Read and write.

Input	Sent in	Type	Required	Meaning
id	path	string (uuid)	yes	The execution id run_workflow answered as executionId.
include_provenance	query	boolean	no	Also answer workflowSnapshot and reproductionManifest. Both are fixed at acceptance, so ask once, not on every poll. Default: False.
project_id	query	string (uuid)	yes	The project this call acts in.

Answer: 200, WorkflowExecutionStatus.

Refused with: 401, 404, 422.

Cancel execution

POST /executions/{id}/cancel

Stop a run that has not finished.

Key: Read and write.

Input	Sent in	Type	Required	Meaning
id	path	string (uuid)	yes	The execution id run_workflow answered as executionId.
project_id	body	string (uuid)	yes	The project this call acts in.

Answer: 200, CancelExecutionResponse.

Refused with: 401, 403, 404, 422.

List execution diagnostics

GET /executions/{id}/diagnostics

Read a run's findings, a page at a time, to see why it failed or what it warned about.

Key: Read only or Read and write.

Input	Sent in	Type	Required	Meaning
id	path	string (uuid)	yes	The execution id run_workflow answered as executionId.
limit	query	integer	no	Page size (default 25, max 100). Default: 25.
offset	query	integer	no	Page offset for the next batch. Default: 0.
project_id	query	string (uuid)	yes	The project this call acts in.
severity	query	error or warning or info	no	Return only findings of this severity; omit for all.

Answer: 200, list of Diagnostic.

Refused with: 401, 404, 422.

Export scenario

POST /scenarios/{id}/export

Prepare a scenario's tables as one file and answer a short-lived signed link to download it.

Key: Read only or Read and write.

Input	Sent in	Type	Required	Meaning
id	path	string (uuid)	yes
project_id	body	string (uuid)	yes	The project this call acts in.

Answer: 200, ArtifactExport.

Refused with: 401, 404, 422.

Refusals

Answer	What it means
401	No integration key was sent, or the key was revoked, replaced by a rotation, or never issued.
403	The key may only read and download, and this operation changes data.
404	Nothing this key reaches has that id. A project outside the key's projects answers exactly as one that does not exist.
422	An input is missing or cannot be read, or the operation refused the request and says why.

Types

Error

A refusal the caller can act on: what is wrong, and, where there is one, what to do about it.

Field	Type	Always	Meaning
error	string	yes
explanation	string	no
remedy	string	no
field	string	no	The input the refusal is about.

Refusal

The operation read the request and refused it, saying why in one diagnostic per problem.

Field	Type	Always	Meaning
diagnostics	list of Diagnostic	yes

ArtifactExport

A prepared, immutable export artifact held for a limited time.

Field	Type	Always	Meaning
download_expires_at	string (date-time) or null	yes
download_url	string or null	yes	A GET-only link to the file, valid until download_expires_at. Null when this deployment has no public object storage.
expires_at	string (date-time)	yes	After this the export can be neither downloaded nor imported.
export_id	string (uuid)	yes
filename	string	yes
guide_url	string	no	The guide section that says, for each agent client, which network setting lets its sandbox fetch download_url's host. If your fetch cannot connect, your environment blocks that host: tell the person the host and this link instead of working around the missing file. Whole, composed by the product, absent when download_url is null.
kind	scenario or data_source	yes
sha256	string	yes
size_bytes	integer	yes
subject_id	string (uuid)	yes	The exported object.

CancelExecutionResponse

Response model for cancelling a workflow execution.

Field	Type	Always	Meaning
executionId	string	yes	This run's execution id. Pass it as execution_id to get (kind execution), cancel (kind execution) and list_execution_diagnostics.
status	string	yes
url	string	no	This run's page, under the workflow it is a run of. Hand this to the person, and report the run by its name with that name as the link — the id is for your next call and is not something to put in front of a reader. The address is whole (scheme, host and path) and the product composed it: never build one yourself out of an id and a host, and if this field is absent, say what was done and name the run without a link.

DashboardRefreshAction

Explicit next step for the requested dashboard. A completed workflow never refreshes saved selections by itself.

Field	Type	Always	Meaning
tool	update	yes
kind	dashboard	yes
refresh_from_execution_id	string	yes
instruction	string	yes

DashboardRefreshDiscovery

Dashboards selecting results of this workflow, whether already refreshed or still pinned. List them with list (kind dashboard) with workflow_id; use update (kind dashboard) with refresh_from_execution_id only for dashboards the analyst wants updated.

Field	Type	Always	Meaning
count	integer	yes
firstDashboard	DashboardRefreshReference or null	yes
refreshAction	DashboardRefreshAction or null	yes

DashboardRefreshReference

A saved dashboard selecting results of this workflow. Refresh only when requested; it may intentionally be historical.

Field	Type	Always	Meaning
id	string	yes	Pass as dashboard_id to get or update (kind dashboard).
name	string	yes
url	string	no	The existing dashboard link, unchanged by refresh.
showsExecution	boolean	yes	Whether this workflow's selected results all come from the reported execution. False means the dashboard link still shows a different run; rerunning did not refresh it.

DataSourceFields

What every data source carries.

Field	Type	Always	Meaning
id	string	yes	The data source id (UUID)
name	string	yes	The data source's name, unique in its project. Its files keep their own names.
updatedAt	string (date-time)	yes
url	string	no	This data source's page, showing what was uploaded. Hand this to the person, and report the data source by its name with that name as the link — the id is for your next call and is not something to put in front of a reader. The address is whole (scheme, host and path) and the product composed it: never build one yourself out of an id and a host, and if this field is absent, say what was done and name the data source without a link.
createdAt	string (date-time)	yes
fileCount	integer	yes	How many files the data source holds.

DataSourceKind

What a data source is for.

tabular sources carry rows and columns and can feed a workflow. document sources are reference prose — assumptions, methodology, results write-ups — stored verbatim and read, never parsed.

tabular or document.

DataSourceListResponse

Response model for listing data sources.

Field	Type	Always	Meaning
dataSources	list of DataSourceResponse	yes
total	integer	yes	Total number of data sources
unfinishedUploads	list of UnfinishedUpload	yes	Uploads that hold a file name in this project and are not files yet, whatever kind asks for. A name one of them holds is refused to every other upload until it finishes or is cancelled.

DataSourceOrigin

How a data source was created.

upload or workflow or assistant.

DataSourceResponse

A data source: a named set of files, shown by its latest file. A workflow's Import Data node reads its latest file or one file of it.

DataSourceFields and object.

DataSourceType

File format of a stored data source, keyed off its extension.

csv or xlsx or xls or md or txt.

DataSourceUploadDestination

Where a data-source upload's file goes: an existing data source, or the name of the one it creates. A scenario bundle's session carries neither.

Field	Type	Always	Meaning
dataSourceId	string (uuid)	no	The data source the file goes into or joined.
newDataSourceName	string	no	Until the file is published: the name of the new data source it creates.

DataSourceUploadLinks

Where a person can finish a data-source upload the agent's own environment could not send, and the guide that says which client setting lets it through. Only a data-source upload carries them.

Field	Type	Always	Meaning
guideUrl	string	no	The guide section that says, for each agent client, which network setting lets its PUTs reach the target host, and how a person uploads in the app instead. Hand it to the person with the refused host when your PUT is blocked. Whole, composed by the product, absent when no address can be given.
url	string	no	Where a person finishes this upload in the app: the project's data-source page until the upload is ready, then the data source itself. Uploading the same file there, under the same name, completes this session. Hand it to the person when your own PUT cannot reach the target host, with the data source's name as the link text. The address is whole (scheme, host and path) and the product composed it: never build one yourself out of an id and a host, and if this field is absent, name the project without a link. A scenario bundle's session carries neither link.

DataSourceUploadSession

An upload session, with the links a data-source upload adds.

UploadSession and DataSourceUploadLinks and DataSourceUploadDestination.

DataSourceUploadSessionInProgress

A data-source upload in progress, with the links a person or agent finishes it by.

UploadSessionInProgress and DataSourceUploadLinks and DataSourceUploadDestination.

DeleteConfirmation

Confirmation that a resource was deleted.

Field	Type	Always	Meaning
deleted	string (uuid)	yes	UUID of the deleted resource.

Diagnostic

One structured finding. Its kind, category, label and evidence contract come from a core or installed package declaration.

Field	Type	Always	Meaning
category	data_quality or configuration or validation or model_feasibility or runtime	yes
cause	string	no	Why, where it is knowable.
details	object	no	Evidence. data_quality findings carry subject naming what in the data is at fault; configuration findings carry none, because there is no data to point at (§3). The kinds with a typed evidence builder have a contract shape here — unmatched_join_keys is subject.{table, keys}. documentation is a repair guide: url opens it for a person; with a node_type, describe (kind node_type) with section = label reads it.
details.subject	list of object	no	Where the finding lives — §4's closed list of refs, and it is a list because every emitter and renderer can name multiple locations. At least one locator field per ref; source_file (the uploaded file a row came from) and source (the record's own semantic origin) are provenance and neither substitutes for the other. Lists truncate at 100 entries with the true count carried alongside. Absent on configuration and runtime findings, which have no data to point at.
details.subject[].column_count	integer	no	How many columns the finding names; columns lists at most the first 100. Written whenever columns is.
details.subject[].columns	list of integer	no	Columns by 1-based position, for findings raised before a column has a usable name.
details.subject[].fields	list of string	no	Columns by name.
details.subject[].key_count	integer	no	How many offending values the finding names; keys lists at most the first 100. Written whenever keys is.
details.subject[].keys	list of string	no	The offending values, when the finding is about values rather than positions.
details.subject[].row_count	integer	no	How many rows the finding names, unreadable row numbers included; rows lists at most the first 100. Written whenever rows is.
details.subject[].rows	list of integer	no	1-based source row numbers.
details.subject[].source	string	no	The record's own semantic origin.
details.subject[].source_file	string	no	The uploaded file a row came from.
details.subject[].table	string	no	The collection: a scenario table, a record type, or an upstream input's label.
kind	string or null	yes	A core or installed package diagnostic identifier.
label	string or null	no	Display label from the emitting diagnostic declaration.
message	string	yes	What is wrong.
node_id	string	no	The workflow node identifier for a graph-shaped finding. Matches the node.id in the workflow document so the app or agent can locate the affected node. Omitted when the finding does not refer to a node.
repair_guidance	string or null	no	Repair guidance from the emitting package or core declaration.
severity	error or warning or info	yes
suggestions	list of string	no	How to repair it.

ExecutionDataSource

A file this run saved, and the data source it joined.

Field	Type	Always	Meaning
id	string (uuid)	yes	The data source. Pass as data_source_id to read the saved table.
name	string	yes	The data source's name.
url	string	no	The data source's page. Hand this to the person, and report the data source by its name with that name as the link — the id is for your next call and is not something to put in front of a reader. The address is whole (scheme, host and path) and the product composed it: never build one yourself out of an id and a host, and if this field is absent, say what was done and name the data source without a link.
fileId	string (uuid)	yes	The file this run saved, its latest unless a later run added another.
fileName	string	yes	That file's name, name [version N].ext.

ExecutionScenario

A scenario this run published.

Field	Type	Always	Meaning
id	string (uuid)	yes	Pass as scenario_id, or in a dashboard's scenario_ids.
name	string	yes	The saved name, including any suffix assigned on repeat runs. It tells a baseline from a variant.
url	string	no	This scenario, open in the Viewer. Hand this to the person, and report the scenario by its name with that name as the link — the id is for your next call and is not something to put in front of a reader. The address is whole (scheme, host and path) and the product composed it: never build one yourself out of an id and a host, and if this field is absent, say what was done and name the scenario without a link.

ExecutionStatus

pending or running or completed or failed or cancelled.

FileFields

What every file of a data source carries, whoever wrote it.

Field	Type	Always	Meaning
columns	list of string	yes	Column names from the file
dataSourceId	string (uuid)	yes	The data source this file belongs to.
id	string	yes	The file id (UUID)
isLatest	boolean	yes	Whether this is the data source's latest file, the one a workflow reading the latest reads.
kind	DataSourceKind	yes	Derived from type — a stored copy could contradict the file.
name	string	yes	The file's name. It is never given to another file
number	integer	yes	The file's place in its data source, counted from 1 in the order files arrived.
origin	DataSourceOrigin	no	How this data source was created. Default: upload.
rowCount	integer	yes	Total table rows; zero for a document
sha256	string	yes	Hex SHA-256 of the stored file's bytes, hashed as they were received — compare it against the file on disk to prove the upload arrived unaltered.
sizeBytes	integer	yes	File size in bytes
type	DataSourceType	yes	File type
updatedAt	string (date-time)	yes	When the file record last changed
uploadedAt	string (date-time)	yes	When the file arrived
uploadedBy	string	yes	The user who uploaded the file or requested the run that wrote it.
url	string	no	The page of the data source this file belongs to. Hand this to the person, and report the data source by its name with that name as the link — the id is for your next call and is not something to put in front of a reader. The address is whole (scheme, host and path) and the product composed it: never build one yourself out of an id and a host, and if this field is absent, say what was done and name the data source without a link.
warnings	list of UploadDiagnostic	no	What reading the uploaded file could not do faithfully: a header that merged or was blank, a workbook sheet that was not read, numbers whose decimal mark the file leaves open. Absent when there is nothing to say. Tell the person before they build on this source.

FileResponse

One file of a data source: bytes received once, under a name no other file in the project has.

FileFields and object.

FindingCounts

Findings per severity.

Field	Type	Always	Meaning
error	integer	yes
info	integer	yes
warning	integer	yes

NodeActivity

The computation currently running inside a workflow node.

Field	Type	Always	Meaning
phase	string	yes
message	string	yes
startedAt	string (date-time)	yes
timeLimitSeconds	number	yes
finishedAt	string (date-time) or null	no	Default: None.
elapsedSeconds	number or null	no	Default: None.

NodeProgress

Progress of a single workflow node during execution

Field	Type	Always	Meaning
nodeId	string	yes
startedAt	string (date-time) or null	no	When the node was first reported running; absent for a node that has not started, and on runs from before it was recorded.
status	ExecutionStatus	yes
activity	NodeActivity or null	no	Default: None.

ProjectListItem

Field	Type	Always	Meaning
briefVersion	integer	yes	Raised by every change to the brief. Send it back as baseBriefVersion when you rewrite the brief.
createdAt	string (date-time)	yes
id	string	yes
isDemo	boolean	no	Default: False.
name	string	yes
querySlug	string or null	no	The former query-link slug, present only when it identifies one project globally. Canonical browser paths use slug. Default: None.
resourceCounts	ResourceCounts or null	no
slug	string	yes
summary	string or null	yes	The brief's opening paragraph, headings skipped, cut at a word and marked with an ellipsis past 280 characters; null when the brief has none.
updatedAt	string (date-time)	yes
uploadUrl	string	no	The project's Data Sources page, where a person drops files into this project. When a file the person gave you cannot reach SCMotif through upload — you cannot compute its SHA-256 or send its bytes from where you run — hand the person this link and wait until list (kind data_source) shows the file; never copy a file's rows into a workflow instead. Whole and composed by the product, like url; absent when url is.
url	string	no	The project's own page, opened with that project selected. Hand this to the person, and report the project by its name with that name as the link — the id is for your next call and is not something to put in front of a reader. The address is whole (scheme, host and path) and the product composed it: never build one yourself out of an id and a host, and if this field is absent, say what was done and name the project without a link.
userId	string	yes

ResourceCounts

Per-project counts of each resource kind.

Field	Type	Always	Meaning
dashboards	integer	no	Default: 0.
dataSources	integer	no	Default: 0.
scenarios	integer	no	Default: 0.
workflows	integer	no	Default: 0.

RunWaiting

Why a run has not started yet. Present only while it is pending: a run the service has taken is no longer waiting for it.

Field	Type	Always	Meaning
aheadOwn	integer	yes	How many of the runs ahead of this one belong to this same account. This is the half of the wait the caller can shorten, by cancelling one of their own.
explanation	string	yes	Why this run is waiting, in one sentence, as the product says it. Say this to the user rather than composing your own: a queued run that says nothing is indistinguishable from a stuck one, and both surfaces state the wait the same way on purpose.
position	integer	yes	This run's place in line for a solver slot, counting itself, so 1 is the run a freeing slot takes next. The service divides itself between accounts rather than starting arrivals in order, so this is not simply how long the run has been queued.
slots	integer	yes	How many solver slots could take this run at all. 0 means none can, and the run will be reported as failed rather than left waiting.

ScenarioImportReady

UploadSessionBase and object.

StorageUsage

How much this account has stored and how much it may. One figure for the whole account, not per project: the limit spans them and survives deleting one. Both halves are counted — uploaded data sources and the result tables runs write — because both are stored.

Field	Type	Always	Meaning
limitBytes	integer	yes	The ceiling for this account. A write that would cross it is refused with a storage_limit_reached diagnostic; nothing already stored is affected, and every read keeps working.
nearLimit	boolean	yes	Whether the account is close enough to its ceiling to say so. Tell the user when this is true — they cannot see it otherwise, and the alternative is finding out at a refusal.
percentUsed	integer	yes	usedBytes as a percentage of limitBytes, 0–100.
remainingBytes	integer	yes	What is left after stored bytes and active upload reservations, never below zero. A file larger than this will be refused.
reservedBytes	integer	yes	Bytes promised to active upload sessions but not yet represented by ready data sources. Reservations prevent concurrent large uploads from overcommitting the account.
scenariosBytes	integer	yes	The half stored as scenario result tables. Runs add to it, and deleting a scenario or its project takes it back.
uploadsBytes	integer	yes	The half stored as uploaded data sources.
usedBytes	integer	yes	Bytes stored, both halves together. This is the number the limit is compared against.

UnfinishedUpload

A file upload that holds its name in the project but is not a file yet.

Field	Type	Always	Meaning
confirmedBytes	integer	yes	Bytes object storage has confirmed so far.
declaredSize	integer	yes
expiresAt	string (date-time)	yes	When a transfer that has not finished lets go of the name by itself. An upload whose file has arrived is being read and does not expire.
id	string (uuid)	yes	The upload_session_id to resume, read or cancel it by.
name	string	yes
state	created or uploading or uploaded or ingesting	yes	created or uploading while the transfer is unfinished, which cancel (kind upload_session) releases; uploaded or ingesting once the file has arrived and is being read.
dataSourceId	string (uuid)	no	The data source the file is going into, when it is an existing one.
newDataSourceName	string	no	The name of the data source the file creates, when it is a new one.

UploadBlankHeaderDiagnostic

A ready upload's warning that one or more header cells were blank.

Field	Type	Always	Meaning
category	data_quality	yes
cause	string	yes
details	object	yes
details.subject	list of object	yes
details.subject[].columns	list of integer	yes
details.subject[].column_count	integer	yes
kind	blank_header	yes
message	string	yes
node_id	file_upload	yes
severity	warning	yes
suggestions	list of string	yes
label	string	yes

UploadConfigurationDiagnostic

A rejected upload's actionable ingestion problem.

Field	Type	Always	Meaning
category	configuration or runtime	yes
details	object	yes
kind	data_source_name_used or storage_limit_reached or uploaded_bytes_mismatch or uploaded_file_too_large or uploaded_file_unreadable	yes
message	string	yes
severity	error	yes
suggestions	list of string	yes

UploadDiagnostic

One exact upload-parser warning shape.

UploadBlankHeaderDiagnostic or UploadHeaderCollapseDiagnostic or UploadUnreadSheetsDiagnostic or UploadUnsettledDecimalMarkDiagnostic.

UploadHeaderCollapseDiagnostic

A ready upload's warning that headers collapsed onto one normalized name.

Field	Type	Always	Meaning
category	data_quality	yes
cause	string	yes
details	object	yes
details.collapsed	list of object	yes
details.collapsed[].column	string	yes
details.collapsed[].headers	list of string	yes
details.subject	list of object	yes
details.subject[].fields	list of string	yes
kind	header_collapse	yes
message	string	yes
node_id	file_upload	yes
severity	warning	yes
suggestions	list of string	yes
label	string	yes

UploadPartTarget

A short-lived, capability-bearing HTTP target for one exact slice of the file. Repeating the same PUT safely replaces that part; callers must never construct or persist the URL.

Field	Type	Always	Meaning
expiresAt	string (date-time)	yes
headers	map of text to string	yes	Headers for this PUT, keyed by name. Non-browser clients must send them, including User-Agent; browser clients retain their own User-Agent and send the remaining headers.
partNumber	integer	yes
url	string	yes	Absolute PUT URL composed by the product. It contains a short-lived capability, not durable storage credentials.

UploadSession

The durable upload record, discriminated by state. Only the ready branch exposes a DataSource id; rejected and ready branches carry the exact diagnostics those states can produce.

UploadSessionInProgress or UploadSessionReady or UploadSessionRejected or UploadSessionExpired or ScenarioImportReady.

UploadSessionBase

Fields every durable upload state returns.

Field	Type	Always	Meaning
confirmedBytes	integer	yes
confirmedParts	integer	yes
declaredSize	integer	yes
expiresAt	string (date-time)	yes	When an incomplete session expires and releases its reservation.
id	string (uuid)	yes
name	string	yes	The file's name.
partSize	integer	yes	Bytes in every part except the final part, which contains the remainder.
projectId	string (uuid)	yes
sha256	string	yes
storage	StorageUsage	yes
targets	list of UploadPartTarget	yes	At most the next bounded batch of missing part targets. Empty while ingesting or terminal; call resume again after sending a full batch.
totalParts	integer	yes

UploadSessionExpired

UploadSessionBase and object.

UploadSessionInProgress

UploadSessionBase and object.

UploadSessionReady

UploadSessionBase and object.

UploadSessionRejected

UploadSessionBase and object.

UploadUnreadSheetsDiagnostic

A ready upload's warning that the workbook holds values on sheets the data source does not: details.sheet is the sheet read, and each details.subject entry's table names a sheet that was not read.

Field	Type	Always	Meaning
category	data_quality	yes
cause	string	yes
details	object	yes
details.sheet	string	yes
details.subject	list of object	yes
details.subject[].table	string	yes
kind	unread_sheets	yes
message	string	yes
node_id	file_upload	yes
severity	warning	yes
suggestions	list of string	yes
label	string	yes

UploadUnsettledDecimalMarkDiagnostic

A ready upload's warning that a ';'-separated file's numbers do not settle whether ',' marks decimals, so they were kept as written: details.value is one such cell, in the column details.subject[0].fields names.

Field	Type	Always	Meaning
category	data_quality	yes
cause	string	yes
details	object	yes
details.subject	list of object	yes
details.subject[].fields	list of string	yes
details.value	string	yes
kind	unsettled_decimal_mark	yes
message	string	yes
node_id	file_upload	yes
severity	warning	yes
suggestions	list of string	yes
label	string	yes

WorkflowExecutionResponse

Response model for workflow execution

Field	Type	Always	Meaning
executionId	string	yes	This run's execution id. Pass it as execution_id to get (kind execution), cancel (kind execution) and list_execution_diagnostics.
status	ExecutionStatus	yes
storage	StorageUsage	yes	The account's storage when this run was enqueued. Poll get (kind execution) for the post-publication figure.
url	string	no	This run's page, under the workflow it is a run of. Hand this to the person, and report the run by its name with that name as the link — the id is for your next call and is not something to put in front of a reader. The address is whole (scheme, host and path) and the product composed it: never build one yourself out of an id and a host, and if this field is absent, say what was done and name the run without a link.
workflowId	string	yes

WorkflowExecutionStatus

Status of a workflow execution

Field	Type	Always	Meaning
completedAt	string (date-time) or null	no
dataSources	list of ExecutionDataSource	yes	Every data source this run wrote, identified without a project-wide list or a name guess.
executionId	string	yes	This run's execution id. Pass it as execution_id to get (kind execution), cancel (kind execution) and list_execution_diagnostics.
findingCounts	FindingCounts	yes	How many findings this run filed, per severity. list_execution_diagnostics reads them, a page at a time and narrowed by severity if asked.
firstFinding	Diagnostic or null	yes	The run's first error, or its first finding of any severity when it filed no error; null when it filed none. The rest are in list_execution_diagnostics.
nodeProgress	list of NodeProgress	no	The durable progress last reported by the currently fenced execution attempt.
reproductionManifest	object or null	no	The immutable inputs, build, packages and effective solver configuration recorded for this run. Present only when include_provenance is true; null for runs created before provenance manifests were recorded.
scenarios	list of ExecutionScenario	yes	Every scenario this run published, in the order it published them, each with its name and page. get (kind scenario) reads one scenario's whole record.
startedAt	string (date-time)	yes
status	ExecutionStatus	yes
storage	StorageUsage	yes	The account's storage at this poll. After completion this includes the run's published result tables; tell the user when nearLimit is true.
url	string	no	This run's page, under the workflow it is a run of. Hand this to the person, and report the run by its name with that name as the link — the id is for your next call and is not something to put in front of a reader. The address is whole (scheme, host and path) and the product composed it: never build one yourself out of an id and a host, and if this field is absent, say what was done and name the run without a link.
waiting	RunWaiting or null	yes	Why this run has not started, while it is still pending; null once it has, or once it is over. Tell the user what it says — a run reported only as pending gives them no way to tell a busy service from a stuck one.
workflowArchived	boolean	yes	Whether the source workflow is archived and unavailable to Run again.
workflowId	string	yes
workflowName	string or null	yes	The workflow name captured when this run was accepted. Null for runs created before execution snapshots were recorded.
workflowSnapshot	object or null	no	The immutable accepted workflow document dispatched for this run, including its canvas layout. Present only when include_provenance is true; null for older runs whose accepted document cannot be reconstructed truthfully.
dashboardRefresh	DashboardRefreshDiscovery or null	no

WorkflowListResponse

Lightweight response model for workflow list (no full node/edge arrays)

Field	Type	Always	Meaning
createdAt	string (date-time)	yes
description	string or null	no
edgeCount	integer	yes
id	string	yes
name	string	yes
nodeCount	integer	yes
updatedAt	string (date-time)	yes
url	string	no	This workflow's page, showing the graph as it now stands. Hand this to the person, and report the workflow by its name with that name as the link — the id is for your next call and is not something to put in front of a reader. The address is whole (scheme, host and path) and the product composed it: never build one yourself out of an id and a host, and if this field is absent, say what was done and name the workflow without a link.
userId	string	yes