quiver-ai-cli

CLI tool for the QuiverAI SVG generation API. Generate SVGs from text prompts and convert raster images to SVG — right from your terminal.

Features

• Text to SVG — describe what you want, get an SVG back
• Image to SVG — convert PNG/JPG/WebP images into clean SVG vectors
• Streaming — watch generation progress in real-time via SSE
• Multiple outputs — generate several SVG variants in one call
• Reference images — attach style references (local files or URLs)
• Zero dependencies — single-file bundle, works with Node.js 18+ or Bun

Install

# npm
npm install -g quiver-ai-cli

# bun
bun install -g quiver-ai-cli

After install, the quiver command is available globally.

Setup

Get your API key from QuiverAI and configure it:

quiver config set api_key sk-xxxxxxxxxxxx

Configuration is saved to ~/.config/.quiver-ai/config.json. You can also use the QUIVER_API_KEY environment variable (takes precedence over config file).

# Optionally set a default model
quiver config set default_model arrow-preview

# View current config
quiver config get

# Show config file path
quiver config path

Usage

Generate SVG from text

# Basic generation
quiver generate "a minimalist rocket icon"

# Save to file
quiver generate -o rocket.svg "a rocket launching into space"

# Stream with progress, specific model
quiver generate -s -m arrow-preview -o icon.svg "a compass icon"

# Generate 3 variants
quiver generate -n 3 -o icons.svg "a navigation arrow"

# With style reference image
quiver generate -r style-ref.png -o logo.svg "a tech startup logo"

# With additional instructions
quiver generate -i "flat design, single color, no gradients" -o flat.svg "a heart icon"

Convert image to SVG

# From local file
quiver vectorize photo.png -o output.svg

# From URL
quiver vectorize https://example.com/logo.png -o logo.svg

# With auto-crop and streaming
quiver vectorize -s --auto-crop image.jpg -o clean.svg

# Resize before vectorizing
quiver vectorize --target-size 512 large-photo.png -o resized.svg

List models

# List all available models
quiver models list

# Get details for a specific model
quiver models get arrow-preview

CLI Reference

quiver config set <key> <value>        Save a configuration value
quiver config get [key]                Show configuration (or a specific key)
quiver config path                     Print the config file path

quiver models list                     List all available AI models
quiver models get <model-id>           Show detailed metadata for a model

quiver generate [options] <prompt>     Generate SVG from a text description
quiver vectorize [options] <image>     Convert a raster image to SVG

Generate options

| Flag | Description | |------|-------------| | -m, --model <model> | Model ID (default: from config or arrow-preview) | | -o, --output <file> | Save SVG to file instead of stdout | | -s, --stream | Stream response via SSE with progress | | -n, --count <n> | Number of SVG variants (default: 1) | | -t, --temperature <t> | Sampling temperature (default: 1) | | -i, --instructions <text> | Additional style guidance | | --max-tokens <n> | Max output tokens | | -r, --reference <url\|file> | Reference image (repeatable) |

Vectorize options

| Flag | Description | |------|-------------| | -m, --model <model> | Model ID (default: from config or arrow-preview) | | -o, --output <file> | Save SVG to file instead of stdout | | -s, --stream | Stream response via SSE with progress | | -t, --temperature <t> | Sampling temperature (default: 1) | | --auto-crop | Auto-crop to dominant subject first | | --target-size <px> | Resize to square of this size before processing | | --max-tokens <n> | Max output tokens |

Development

# Install dependencies
bun install

# Run in dev mode (directly from TypeScript)
bun run dev

# Build for Node.js (outputs dist/index.js)
bun run build

# Build standalone binary (outputs dist/quiver)
bun run build:binary

# Run built version
bun run start

License

MIT