Model Context Protocol server for Roshan AI Alefba OCR with multilingual document understanding, exports, async processing, and support for multiple self-hosted instances.
| Tool name | Description | Destructive? |
|---|---|---|
| alefba_read_document | Read (OCR) a document/image from a URL; sync (wait=true) or async (task_ids). | ✓ no |
| alefba_read_document_upload | Upload a local file and read (OCR) it via multipart. | ✓ no |
| alefba_get_result | Fetch/poll an async read result by task_id. | ✓ no |
| alefba_read_pages | Read specific document pages given as URLs with @page=N. | ✓ no |
| alefba_document_status | Report per-document processing progress. | ✓ no |
| alefba_document_pages | List the page URLs of a document. | ✓ no |
| alefba_download_word | Download the analyzed document as Word (.docx). | ✓ no |
| alefba_download_excel | Download the analyzed document as Excel (.xlsx); requires type=excel. | ✓ no |
| alefba_download_pdf | Download the analyzed document as a searchable PDF. | ✓ no |
| alefba_delete_document⚠ | Delete a document and its results. | ⚠ yes |
| alefba_read_document_callback | Process a document and receive the result via webhook callback. | ✓ no |
| healthcheck | Check that an Alefba instance is up and ready. | ✓ no |
| list_instances | List configured Alefba instances (names and base URLs only). | ✓ no |
| roshan_alefba_docs | Return documentation about Alefba and these tools. | ✓ no |