Quiver Block

The Quiver block generates clean, scalable SVG graphics from text prompts or existing images using AI. Reach for it when you need vector artwork, icons, or illustrations produced on the fly within a workflow, with support for multiple generation models and optional reference images.

Overview

When to Use

• Generate SVG icons or illustrations from a natural-language description.
• Vectorize an existing raster image (PNG/JPG) into a scalable SVG.
• Produce multiple SVG variants of a concept in a single step.
• Steer output with reference images and detailed style instructions.
• Discover which Quiver models are available before generating.
• Tune generation behavior (temperature, top-p, presence penalty, output length) for deterministic or more varied results.

Configuration

Operation

Dropdown that selects what the block does. Defaults to text_to_svg.

API Key

Required. Your Quiver API key, stored as a password field. Used by all three operations.

Model

Dropdown to select the generation model. Shown for text_to_svg and image_to_svg. Defaults to quiver-mini.

Prompt

Long-text input. Required for text_to_svg. A natural-language description of the SVG to generate.

Instructions

Long-text input. Optional style or formatting guidance for the text_to_svg operation (e.g., "flat, minimal style with rounded corners").

Reference Images

File-upload field (basic mode, multiple files allowed). Provide up to 4 reference images to guide generation. Shown for text_to_svg.

Reference Image URLs

Short-text input (advanced mode). Comma-separated image URLs used as references when operating in advanced mode. Shown for text_to_svg.

Number of Variants

Short-text input (advanced mode). How many SVG variants to generate (1–16, default 1). Shown for text_to_svg.

Image File

File-upload field (basic mode, single file, required). The raster image to vectorize. Shown for image_to_svg.

Image URL

Short-text input (advanced mode, required). URL of the raster image to vectorize. Shown for image_to_svg.

Auto Crop

Dropdown (advanced mode). Whether to automatically crop the image before vectorizing. Shown for image_to_svg.

Target Size (px)

Short-text input (advanced mode). Square resize target in pixels (128–4096). Shown for image_to_svg.

Temperature

Short-text input (advanced mode). Sampling temperature (0–2, default 1.0). Available for all operations.

Top P

Short-text input (advanced mode). Nucleus sampling probability (0–1, default 1.0). Available for all operations.

Max Output Tokens

Short-text input (advanced mode). Maximum output tokens (1–131072, default 65536). Available for all operations.

Presence Penalty

Short-text input (advanced mode). Token penalty for tokens that already appear in output (−2 to 2, default 0). Available for all operations.

Inputs & Outputs

• Inputs:

  • operation (string) — Operation to perform (text_to_svg, image_to_svg, or list_models)
  • apiKey (string) — Quiver API key
  • model (string) — Model to use for generation
  • prompt (string) — Text prompt for SVG generation (text_to_svg only)
  • instructions (string) — Style instructions for the SVG (text_to_svg only)
  • referenceFiles (json) — Reference image files uploaded directly (text_to_svg, basic mode)
  • referenceInput (string) — Reference image URLs, comma-separated (text_to_svg, advanced mode)
  • n (number) — Number of SVG variants to generate (text_to_svg only)
  • imageFile (json) — Image file to vectorize (image_to_svg, basic mode)
  • imageInput (string) — Image URL to vectorize (image_to_svg, advanced mode)
  • autoCrop (string) — Auto crop the image before processing (image_to_svg only)
  • targetSize (number) — Target size in pixels for resizing (image_to_svg only)
  • temperature (number) — Generation temperature
  • topP (number) — Top-p sampling parameter
  • maxOutputTokens (number) — Maximum output tokens
  • presencePenalty (number) — Presence penalty

• Outputs:

  • file (json) — Generated SVG file (first result)
  • files (json) — Array of all generated SVG files when n > 1
  • svgContent (string) — SVG content as a raw markup string
  • id (string) — Generation request ID
  • usage (json) — Token usage statistics (totalTokens, inputTokens, outputTokens)
  • models (json) — List of available QuiverAI models (list_models only)

Tools

Quiver Text to SVG (quiver_text_to_svg) — Calls the QuiverAI API to generate one or more SVG images from a text prompt. Supports reference images, style instructions, and sampling parameters.

Quiver Image to SVG (quiver_image_to_svg) — Converts a raster image (supplied as a file or URL) into vector SVG format using QuiverAI. Supports auto-crop, target-size resizing, and sampling parameters.

Quiver List Models (quiver_list_models) — Retrieves all available QuiverAI models from the /v1/models endpoint, returning metadata including supported operations, input/output modalities, and context lengths.

YAML Example

quiver_1:
  type: quiver
  name: "Quiver"
  inputs:
    operation: "text_to_svg"
    apiKey: "{{QUIVER_API_KEY}}"
    model: "quiver-mini"
    prompt: "{{start.input}}"
    instructions: "Flat, minimal style with rounded corners"
    n: 1
  connections:
    outgoing:
      - target: next-block-id