- description
- Generates AI-powered descriptions for entities using Gemini. Supports text properties, relationships, and file content (images, PDFs, documents up to 50MB). Streams files directly to Gemini Files API without memory buffering.
- endpoint
- https://describe.nick-chimicles-professional.workers.dev
- endpoint_verified_at
- 2026-02-23T19:50:32.537Z
- input_schema
- description
- Parameters passed via the rhiza workflow input field to customize description generation. Properties map directly to keys in request.input.
- properties
- batch_size
- default
- 100
- description
- Number of entities to fetch per batch-get call.
- maximum
- 500
- minimum
- 1
- required
- false
- type
- integer
- content_keys
- default
- all keys
- description
- Specific content keys to include. If omitted, all content keys are included.
- examples
- page_001.jpeg
- page_002.jpeg
- required
- false
- type
- string[]
- context_window_tokens
- default
- 128000
- description
- Model context window size for token budget calculation.
- required
- false
- type
- integer
- custom_instructions
- description
- Additional instructions appended to the system prompt for customizing output.
- required
- false
- type
- string
- focus
- description
- What aspect of the entity to emphasize in the description.
- required
- false
- type
- string
- include_content
- default
- true
- description
- Whether to upload and include file content from properties.content in the prompt.
- required
- false
- type
- boolean
- max_content_size
- default
- 52428800
- description
- Maximum file size in bytes to upload. Larger files are skipped with 'too_large' status.
- notes
- Default is 50MB. Set lower to skip large files and rely on text context only.
- required
- false
- type
- integer
- max_output_tokens
- default
- 8000
- description
- Tokens reserved for model output.
- required
- false
- type
- integer
- max_relationships
- default
- 1000
- description
- Maximum number of relationships to fetch for context.
- maximum
- 10000
- minimum
- 0
- required
- false
- type
- integer
- predicates
- default
- []
- description
- Filter to only fetch relationships with these predicates. Empty array means all predicates.
- required
- false
- type
- string[]
- safety_margin
- default
- 0.8
- description
- Buffer ratio to avoid hitting token limits. 0.8 means use 80% of available budget.
- maximum
- 1
- minimum
- 0.5
- required
- false
- type
- number
- style
- default
- detailed
- description
- Writing style for the generated description.
- enum
- concise
- detailed
- academic
- casual
- required
- false
- type
- string
- update_label
- default
- false
- description
- If true, also generates and updates a concise label (2-5 words) for the entity.
- required
- false
- type
- boolean
- status
- active