Alkymi - The Enterprise AI Platform for Private Markets Data

List Flattened Records

List flattened records from Viaduct with optional filtering and pagination. Each record includes extracted field values inline in a 'data' array alongside execution, document, dataset, pipeline, and schema identifiers. IMPORTANT: Always prefer targeted filters over paginating through results: - Use fund, lp, or borrower to filter by primary object — this is the most efficient way to find records associated with a specific entity. - Use execution_uuid to get all records from a specific docume

List Executions

List executions (uploaded documents) from Alkymi with optional filtering and pagination. Each execution represents a document submitted for extraction. IMPORTANT: Always prefer targeted filters over paginating through results: - Use filename or search to locate a document by name — far more efficient than scanning pages. - Use fund, lp, or borrower to filter by primary object when looking for documents associated with a specific entity. - Fall back to pagination only when no targeted filte

List Tags

List tags from Alkymi with optional filtering and pagination. Tags are labels that can be applied to executions. Use this to discover available tag UUIDs before setting them via update_execution.

List Pipelines

List available pipelines from Alkymi with pagination. Each pipeline represents a document processing workflow. The schemas and record_validators fields are stripped from the response to keep it concise. Use pipeline names when uploading documents via upload_document (targets parameter) or when filtering records via list_records_flat.

Get Pipeline Config

Export a pipeline's full configuration as a JSON array of Django fixture objects. The returned array can be passed directly to import_pipeline_config to copy or restore the pipeline on any tenant. To duplicate the pipeline (create a new one instead of updating), replace all UUIDs in the payload before importing.

Import Pipeline Config

Import a pipeline configuration from a JSON array of Django fixture objects. Accepts the payload returned by get_pipeline_config. The pipeline is identified by the UUID embedded in the payload: - If a pipeline with that UUID already exists in the tenant, it is updated in place. - If no matching pipeline exists, a new one is created. IMPORTANT — when creating a new pipeline, you MUST set enabled: true on every fixture object in the payload that has an enabled field. This includes the Pipeline

Update Execution

Update an execution's reviewed or removed status. Setting reviewed=true marks the execution as human-reviewed and records the reviewer and timestamp. Setting reviewed=false clears the review. Setting removed=true soft-deletes the execution; removed=false restores it.

Update Record

Update the field values of an existing record. Pass a dict of field names to new values under 'data'. Keys must match the schema's export field names. Each value must be an object with a "text" key, e.g. {"Amount": {"text": "1500.00"}, "Currency": {"text": "USD"}}.

Remove Record

Permanently delete a record by UUID. This action is irreversible. The record and all its field values will be removed from the system.

Upload Document

Upload a file to Alkymi for extraction. Two ways to provide the file: - file_path: absolute path to a local file on the server filesystem. - file_name: name of a file already uploaded to the server via the file_manager tool. Use file_manager to upload the file first, then pass its name here. Submits a new execution and returns the created execution object on success. IMPORTANT: After uploading, track extraction progress by reading the alkymi://executions/{uuid} resource (substituting the

List Files

List all uploaded files with metadata.

Read File

Read an uploaded file's contents by name.

File Manager

Upload and manage files. Drop files here to send them to the server.