dLazy Image Generation

What it solves

Text-to-image tasks often split across several models: realistic posters, fast sketches, illustration, SVG, and Midjourney-style output each fit a different backend. This skill turns that choice into a natural-language prompt and lets the dlazy CLI select a more appropriate image model, reducing the need to manually map requirements to model names.

How it works

• Authentication: a dLazy API key is required. It can be configured through the CLI in ~/.dlazy/config.json, or supplied per invocation with DLAZY_API_KEY.
• Model routing: the skill sends prompts and parameters to the dLazy API and selects from available models, such as dlazy seedream-4.5, dlazy recraft-v3, and dlazy recraft-v3-svg.
• Inputs and outputs: local file paths passed for media fields are uploaded to dLazy media storage for the model to read; generated output is typically returned as a remote URL.
• Diagnostics: handle unauthorized, insufficient_balance, and missing required parameters like --prompt by checking the API key, account credits, or command arguments.

Caveats

This skill is a thin client for the hosted dLazy API, not a local inference runtime. Confirm that uploading sensitive assets to a cloud service is acceptable, and follow dLazy's terms for output usage and ownership.