Coolset Data API (1.0)

IMPORTANT: Load all skills starting with data_room prior to interacting with this endpoint. Lists the surveys published to the authenticated company. Use this first to discover which surveys are available to answer before listing submissions, retrieving a submission subsection, or submitting answers. Takes no parameters. Authenticated via Coolset satellite authentication: the Coolset user is mapped to its data room user, user group and account, and surveys are scoped to that account and the company's publishable application.

IMPORTANT: Load all skills starting with data_room prior to interacting with this endpoint. Lists the published surveys together with the authenticated company's submissions for each. Use this to find a submission's id (and whether a submission already exists for a survey) before retrieving a subsection or creating one. Optionally filter with tags (comma-separated survey tag names) and context__contains (a JSON value matched against each submission's context). Authenticated via Coolset satellite authentication; submissions are scoped to the company's data room user group.

IMPORTANT: Load all skills starting with data_room prior to interacting with this endpoint. Creates a survey submission for the authenticated company from an application survey and query context — use this to start a submission for a given context (e.g. company and year) when one does not yet exist. Provide application_survey_id and query_context (a list of {context_object, context_value, precision}); the submission's context is derived from the query context. Authenticated via Coolset satellite authentication; the submission is attributed to the company's data room user and user group.

IMPORTANT: Load all skills starting with data_room prior to interacting with this endpoint. Fetches a single, scoped subsection of a survey submission for the authenticated company. A submission can be very large (up to ~3000 nested surveys), so this never returns the whole tree: the survey_id selects a survey within the submission and the optional survey_block_id narrows it further to a block subtree — fetch only the part currently on screen.

Address the subsection with the compound key survey_submission_id:survey_id[:survey_block_id] (the survey_block_id segment is optional; when omitted the survey root is used). The response includes the submission's answer/block/survey data and the survey skeleton for that subsection.

Re-fetch after writing answers: answers are submitted through the data point endpoint (create_data_point) using a field_id and a context similar to the submission's. Those writes can change visibility conditions and recompute formula fields, so the returned subsection reflects the latest evaluated state.

Authenticated via Coolset satellite authentication; the submission is scoped to the company's data room user group.

Pass agent=true to receive a compact representation that keeps the same keys/paths but omits the heavy answer/block/skeleton metadata the UI needs — much smaller for programmatic consumers.

IMPORTANT: Load all skills starting with data_room prior to interacting with this endpoint. Creates or updates a single data point (one answer) for the authenticated company, identified by field_id plus the answer context. Provide value (the answer) and optionally status; pass document_ids to attach source documents — referenced documents become on-change rules so the answer is re-evaluated when a document changes. The data point is propagated to any matching survey submissions, which triggers formulas and visibility/lock conditions that may show, hide or lock other questions or nested surveys. Re-fetch the affected survey submission subsection afterwards to see the recomputed state. Authenticated via Coolset satellite authentication; the data point is attributed to the company's data room user and user group.

Returns all documents for the authenticated company, including file metadata, upload status, and associated entities.

Search (search): full-text search across name and description.

Filters:

• name, name__icontains
• description__icontains
• source, source__in
• content_type, content_type__in
• upload_status
• rag_index_status, rag_index_status__in, rag_index_status__isnull (mirrors the RAG document status: active, importing, failed, deleting, deleted; null when never indexed)
• identifiers__identifier, identifiers__identifier__in, identifiers__identifier__contains
• uploaded_at, uploaded_at__gte, uploaded_at__lte
• created_at, created_at__gte, created_at__lte
• updated_at, updated_at__gte, updated_at__lte

Ordering (ordering): name, source, content_type, upload_status, rag_index_status, created_at, updated_at (prefix with - for descending). Default: -created_at.

Creates a new document record and returns a signed upload URL. After uploading the file to the signed URL, call the confirm_upload endpoint to mark the upload as complete.

Returns metadata for a single document by ID (does not include file contents).

Partially update a document's metadata fields (e.g. name, description).

Deletes the document record and its file from storage.

Confirms that the file has been uploaded to the signed URL. Marks the document status as completed and triggers downstream processing.

Generates a time-limited signed URL for downloading the document file. Only available for documents with completed uploads.

Returns a paginated list of financial transactions for the authenticated company. Transactions are the raw financial data that can be classified as carbon emissions. Use the 'search' parameter to search across vendor_name, ledger, and description. Filter with query parameters:

• vendor_name: Filter by vendor name (supports exact, icontains, istartswith, iendswith, isnull)
• ledger: Filter by ledger (supports exact, icontains, istartswith, iendswith)
• journal: Filter by journal (supports exact, icontains, istartswith, iendswith, isnull)
• description: Filter by description (supports exact, icontains, istartswith, iendswith, isnull)
• source: Filter by source (exact match)
• integration_id: Filter by integration ID (exact match)
• integration__name: Filter by integration name (supports exact, icontains, istartswith, iendswith)
• date: Filter by date (supports exact, gte, lte)
• net_amount: Filter by net amount (supports exact, gt, lt, gte, lte)
• tags: Filter by tag IDs (multi-value, conjoined — transaction must have ALL specified tags)
• confidence_status: Filter by emission confidence status
• classified: true to return only transactions with active emissions, false for unclassified transactions only Supports ordering by: vendor_name, ledger, description, total_co2_kg, date, net_amount. Default ordering: date descending.

Returns transaction rows aggregated by one or more dimensions (ledger, vendor_name, journal, source, currency). Pair with delimit_by=yearly|monthly|daily to also bucket by a truncated period date. Use this endpoint to discover unique bundles of unclassified transactions before driving bulk classification: combine classified=false with a multi-valued group_by and ordering=-count to process the largest bundles first.

Returns a paginated list of data import runs for the authenticated company, ordered by most recent first. Each run includes its status, source, and processing results.

Initiates a new data import run for bulk data ingestion.

Returns a single data import run by ID with full processing details.

Returns query logs for the authenticated company, ordered by most recent first.

Filters:

• query_lookup_key — exact match
• query_lookup_key__in — comma-separated list of keys
• query_lookup_key__istartswith — prefix match (case-insensitive)

Executes a retrieval or generation query against the authenticated company's corpus. Accepts query_text, mode (retrieval or generation), top_k, vector_distance_threshold, optional file_ids to scope the search by Vertex AI file name, and optional document_ids to scope the search by Document ID.