Skip to main content

Document Management

Document management is a core feature of Knowledge Base, supporting multiple document sources and formats.

πŸ“₯ Adding Documents​

Supported Document Sources​

SourceDescription
File UploadUpload files from local computer
Text PastePaste text content directly
External TableImport from DingTalk/Feishu tables
Web ScrapingScrape content from URL

File Upload​

  1. Click Add Document β†’ Upload File
  2. Select file(s) from your computer
  3. Configure chunking settings (optional)
  4. Click Upload

Supported formats:

  • .txt - Plain text files
  • .md - Markdown files
  • .pdf - PDF documents
  • .doc, .docx - Word documents

Text Paste​

  1. Click Add Document β†’ Paste Text
  2. Enter document title
  3. Paste or type content
  4. Click Save

Suitable for quickly creating small documents.

External Table​

  1. Click Add Document β†’ External Table
  2. Enter table URL (DingTalk/Feishu)
  3. Configure sync settings
  4. Click Import

Supports importing data from online table services.

Web Scraping​

  1. Click Add Document β†’ Web URL
  2. Enter the webpage URL
  3. System scrapes and processes content
  4. Click Import

Web documents support re-scraping for updates. When webpage content changes, use the refresh feature to get the latest content.

πŸ“‹ Document List​

List Features​

  • Search: Search documents by name
  • Sort: Sort by name, size, or date
  • Filter: Filter documents by status
  • Folder filter: Open APIs can list documents directly under a specific folder by folder_id

Document Status​

StatusDescription
EnabledDocument is indexed and searchable
DisabledDocument exists but excluded from search
QueuedDocument waiting to be processed
Pending ConversionDocument waiting for format conversion (PDF/PPTX etc.)
ConvertingDocument being converted to Markdown
IndexingDocument being indexed for RAG
ErrorIndexing failed

πŸ“ Folder Management​

Knowledge base documents can be organized by folders. Folders are document metadata, not attachment metadata; the generic attachment upload API does not accept folder_id.

Folder ID semantics:

  • folder_id=0 means the root folder.
  • Omitting folder_id means no folder filter.
  • When listing documents by folder, the folder filter returns only direct documents in that folder and does not recursively include subfolders.
  • For the complete open APIs for creating, moving, deleting folders, and moving documents, see Knowledge Open API.

πŸ”Ž Scoped Open Search​

Open search can be scoped to folders or specific documents. This is useful when you only want to search inside a topic folder, root-level documents, or a selected document set.

Parameter rules:

  • Whole-knowledge-base search: omit both folder_ids and document_ids.
  • Root-folder search: pass folder_ids: [0].
  • Subfolder search: pass folder_ids: [10] and use include_subfolders to control whether descendants are included.
  • When both folder_ids and document_ids are provided, the scope is their union.
  • folder_ids=[] or document_ids=[] is invalid. Empty arrays must not be used to mean whole knowledge base.
  • If the specified scope contains no documents, the API returns empty results and does not fall back to whole-knowledge-base search.
  • For request examples and the full parameter reference, see Knowledge Open API.

✏️ Management Operations​

Basic Operations​

OperationDescription
View DetailsView document content and metadata
EditModify document name and settings
Enable/DisableToggle document search participation
Re-indexReprocess document with new settings
DeleteRemove document permanently
View ChunksInspect how document was split

Batch Operations​

Support multi-select for batch operations:

  1. Use checkboxes to select multiple documents
  2. Click Select All to select all documents
  3. Click Batch Delete to delete selected documents

πŸ“Œ Document Selection in Notebook Mode​

In Notebook mode, the document list supports selection features:

Selection Features​

  • Select Specific Documents: Check documents to include in context
  • Select All / Deselect All: Quickly select or deselect all documents
  • Auto-selection: Newly uploaded documents are automatically selected

Context Injection​

Selected documents are provided as context to the AI during conversations, helping the AI better understand and answer questions.

πŸ” Document Editing​

Editable Content​

  • Document Name: Modify the display name
  • Chunking Settings: Adjust document chunking strategy
  • Enable Status: Control whether document participates in retrieval

Editing Limitations​

  • Source Type: Cannot change document source type
  • File Content: File-type document content cannot be directly edited
  • Table URL: External table URLs cannot be directly modified

πŸ”„ Web Document Refresh​

Web documents support re-scraping:

  1. Find the web document in the document list
  2. Click the Refresh button
  3. System will re-scrape the webpage content
  4. Updated content will be automatically re-indexed

Suitable for tracking frequently updated web content.

πŸ’‘ Best Practices​

Document Organization​

PracticeDescription
Meaningful namesUse descriptive document names
Consistent formatStandardize document formatting
Regular updatesRe-index when documents change
Clean contentRemove irrelevant headers/footers

Document Size​

  • Single file recommended not to exceed 50MB
  • Large documents can be split into multiple smaller documents
  • Text documents are easier to process than scanned PDFs

πŸ”— Related Documentation​