Skip to main content
This document covers import_from_sharepoint (display name “SharePoint Import”), the tool that copies one file from a SharePoint document library into platform storage. The returned storage_url can be passed to extraction, CSV, code, and export tools.

​
When to use

Use this tool when an agent needs one known SharePoint file during a task. Use a file_import trigger instead when a folder must be watched continuously for new files. The import is a point-in-time copy. Changes made in SharePoint afterward do not update the stored copy.

​
Authentication and enablement

The tool uses the workspace’s SharePoint connector. Its OAuth token is injected by the platform and not shown as a per-run input. Outbound Microsoft Graph calls are audited. The tool registers only when SHAREPOINT_CLIENT_ID is configured. Off by default — enable it in the agent’s Tools panel.

​
Inputs

Choose exactly one file selector:
  • sharing_url: SharePoint file sharing URL. Do not also provide site_id or drive_id.
  • item_id: Microsoft Graph drive-item ID.
  • file_path: Exact slash-separated path inside the document library.
For item_id and file_path, also provide:
  • drive_id (preferred): Exact document-library drive ID; or
  • site_id: SharePoint site ID. The site’s default document library is used when drive_id is omitted.
Stable Graph IDs returned by this tool can be reused by workbook and export operations.

​
Idempotency

The tool derives an idempotency key from the authoritative site_id, drive_id, and item_id. Repeating the same resolved import within one task returns the existing platform file with duplicate: true instead of uploading another copy. Idempotency is scoped to the task and workspace; it does not silently reuse another tenant’s artifact.

​
Size and file restrictions

  • Maximum file size: 250 MB.
  • Folders and non-downloadable items are rejected.
  • Empty files are rejected.
  • The limit is checked from Graph metadata and again while streaming the download.
Downloads are streamed through a temporary file rather than buffered entirely in memory.

​
Output

The structured result contains:
  • file_id, file_name, storage_url, mime_type, and file_size
  • Authoritative site_id, drive_id, and item_id
  • Source sharepoint_url
  • idempotency_key
  • duplicate
On React tasks, storage_url may be returned as a ${VAR_N} reference so downstream tools can consume it without exposing the raw storage URL in later steps.

​
Errors

  • Missing or multiple selectors fail validation before a Graph request.
  • item_id or file_path without drive_id or site_id is rejected.
  • Sharing URLs that do not resolve to authoritative site, drive, and item IDs are rejected.
  • Folder, empty-file, oversized-file, download, storage, and file-registration failures return structured tool errors.