klados

KG Extractor

01KJ60XQBHJ0GBGTP9X8HXAPPM

Properties

accepts
cardinality
one
types
  • text_chunk
  • document
  • *
actions_required
  • entity:view
  • entity:create
  • entity:update
description
Extracts named entities and relationships from text content using Gemini. Requires entities with a 'text' property (or text file attachment) — cannot process raw files like PDFs or images. Route files through OCR or PDF extraction first.
endpoint
https://kg-extractor-v2.nick-chimicles-professional.workers.dev
endpoint_verified_at
2026-02-23T19:52:00.614Z
input_schema
description
Optional parameters to customize extraction behavior. Properties map directly to keys in request.input.
properties
instructions
description
Custom instructions appended to the extraction prompt. Use to guide what types of entities to focus on, what to skip, or how to interpret domain-specific content. Example: 'Focus on legal entities and contractual relationships. Skip generic organizational mentions.'
required
false
type
string
target_entity_count
description
Override the default target of ~10-20 entities per chunk. Higher values extract more granular entities; lower values focus on only the most significant ones.
maximum
100
minimum
1
required
false
type
integer
match
or
  • exists
    true
    path
    properties.text
  • in
    • text_chunk
    • document
    path
    type
produces
cardinality
many
types
  • person
  • creature
  • ship
  • place
  • group
  • object
  • concept
  • organization
  • event
  • entity
status
active
KG Extractor | Arke