DocRouter Knowledge Base icon

DocRouter Knowledge Base

Manage knowledge bases and run search/chat in DocRouter.ai

Actions11

Create

AI-generated

Summary

Create a new knowledge base in DocRouter.ai by specifying its name and optional settings such as description, tag IDs for auto-indexing, chunking parameters, embedding model, neighbor coalescing, and reconciliation options.

Inputs

  • Name (required) — Human-readable name for the knowledge base.
  • Description — Optional description for the knowledge base.
  • Tag IDs — Comma-separated tag IDs used for automatic indexing within the knowledge base.
  • Chunker Type — The chunker type to divide text into chunks; defaults to 'recursive'.
  • Chunk Size — Target number of tokens per chunk, with a minimum of 1; defaults to 512.
  • Chunk Overlap — Number of overlapping tokens between chunks, minimum 0; defaults to 128.
  • Embedding Model — LiteLLM embedding model to use; defaults to 'text-embedding-3-small'.
  • Coalesce Neighbors — Number of neighboring chunks to include together (0-5); defaults to 0.
  • Reconcile Enabled — Boolean to enable periodic automatic reconciliation for the knowledge base; defaults to false.
  • Reconcile Interval (seconds) — Interval in seconds for reconciliation to run, required if reconciliation is enabled and must be at least 60 seconds.

Output shape

a single record containing the newly created knowledge base's details as JSON

The output includes properties reflecting the created knowledge base, such as its ID and configuration fields. Only one output item is returned per node execution for the create operation.

Examples

Example 1: Create a knowledge base named 'Support Docs' with a description and tag IDs.

Set Name to 'Support Docs', Description to a brief text, and Tag IDs to comma-separated values like 'tag1,tag2'.

Links

Discussion