CLI Usage
Command-line interface reference for automation and scripting.
Asset Tap includes a full-featured command-line interface for automation, scripting, and headless generation.
Installation
The CLI is included with all installers, or available as a standalone download. See the Installation page for platform-specific instructions.
macOS -- The CLI is bundled inside the app. Create a symlink to use it from the terminal:
sudo ln -sf "/Applications/Asset Tap.app/Contents/MacOS/asset-tap" /usr/local/bin/asset-tap
Linux (.deb) -- The CLI is installed to /usr/bin/asset-tap automatically.
Windows -- The CLI is available after install if the installer adds the install directory to PATH.
API Key Configuration
The CLI needs an API key from your provider (e.g., fal.ai). There are two ways to configure it:
Option 1: Environment variable (recommended for CLI)
export FAL_KEY=your_key_here
Add this to your shell profile (~/.zshrc, ~/.bashrc) to persist across sessions.
Option 2: GUI settings (shared automatically)
If you've configured your API key in the Asset Tap GUI (Settings > API Keys), the CLI picks it up automatically -- both share the same settings file.
Basic Usage
# Generate a 3D model from a text prompt
asset-tap --yes "a wooden treasure chest"
# Short form
asset-tap -y "a dragon"
The --yes / -y flag skips confirmation prompts and runs with defaults.
Specifying Provider and Models
# Use a specific provider
asset-tap -p fal.ai -y "a spaceship"
# Choose specific models
asset-tap -p fal.ai --image-model nano-banana-2 --3d-model trellis-2 -y "a robot"
# Use premium image model
asset-tap -p fal.ai --image-model nano-banana-pro -y "a detailed castle"
# Use original Nano Banana
asset-tap -p fal.ai --image-model nano-banana -y "a simple cube"Using an Existing Image
Skip the text-to-image step by providing your own image:
# Convert an existing image to 3D
asset-tap --yes --image "photo.png"
# With a specific 3D model
asset-tap --yes --image "photo.png" --3d-model trellis-2Templates
Use prompt templates to structure your input with predefined formats:
# List available models and templates
asset-tap --list
# Use a template (your prompt becomes the template's description variable)
asset-tap -t humanoid -y "a brave knight with a glowing sword"
# Inspect a template's syntax and preview
asset-tap --inspect-template humanoidListing Providers and Models
# List all available providers and their models
asset-tap --list-providers
# List models and templates
asset-tap --listOutput
Generated assets are saved to timestamped directories. See Bundle Structure for the full output format.
# Use a custom output directory
asset-tap -o ~/my-assets -y "a treasure chest"output/
└── 2024-12-29_153045/
├── bundle.json # Metadata
├── image.png # Generated image
├── model.glb # 3D model
├── model.fbx # FBX (if Blender installed)
└── textures/ # Extracted texturesExporting Bundles
# Export a bundle directory as a zip archive
asset-tap --export-bundle output/2024-12-29_153045FBX Conversion
By default, Asset Tap converts GLB models to FBX if Blender is installed.
# Skip FBX conversion (GLB output only)
asset-tap --no-fbx -y "a robot"
# Convert existing GLB files to FBX (no API calls)
asset-tap --convert-onlyImage Approval
In interactive mode, you can require approval of the generated image before proceeding to 3D generation:
# Require image approval before 3D conversion
asset-tap --approve "a detailed spaceship"Texture Conversion
Some 3D models contain WebP textures that aren't supported by all tools. Convert them to PNG:
# Convert WebP textures in existing GLB files to PNG
asset-tap --convert-webpMock Mode
Test the full pipeline without consuming API credits. Mock mode redirects all API requests to a local server that returns synthetic data (test images and 3D models). This verifies that your configuration, pipeline, and file output work correctly.
# Instant mock responses
asset-tap --mock -y "test prompt"
# With realistic delays
asset-tap --mock --mock-delay -y "test prompt"
Note: Mock mode returns generic responses regardless of which provider or model you select. It validates the pipeline and configuration plumbing, not provider-specific response parsing. To verify a custom provider's response format, test against the real API.
Complete Flag Reference
| Flag | Short | Description |
|---|---|---|
--yes | -y | Auto-confirm all prompts (non-interactive mode) |
--provider | -p | Provider to use (e.g., fal.ai) |
--image-model | Image generation model | |
--3d-model | 3D generation model | |
--image | Skip image generation, use existing image (local path or URL) | |
--template | -t | Use a prompt template |
--output | -o | Output directory for generated assets |
--list | List available models and templates | |
--list-providers | List available providers and their models | |
--inspect-template | Inspect a template's syntax and preview | |
--no-fbx | Skip FBX conversion (GLB only) | |
--convert-only | Convert existing GLB files to FBX (no API calls) | |
--convert-webp | Convert WebP textures in GLB files to PNG | |
--approve | Require image approval before 3D generation | |
--export-bundle | Export a bundle directory as a zip archive | |
--mock | Run in mock mode (no API costs) | |
--mock-delay | Add realistic delays in mock mode |