dLazy Midjourney Image Generation

Problem

When an agent needs Midjourney image output, hand-rolling HTTP calls exposes API key handling, parameter formatting, async polling, and error recovery. mj-imagine narrows that to one CLI invocation: send a prompt and parameters, then receive image results or task status.

How It Works

• Text-to-image: start Midjourney generation with dlazy mj.imagine, making painting requests easier to script and replay.
• API and credentials: requests use the dLazy hosted API and require a dLazy API key; keys are scoped to the organization and can be rotated or revoked, or supplied per run via DLAZY_API_KEY.
• Parameters and output: pipe references such as -, @N, and @stdin:path help connect upstream output to downstream fields; async tasks return a generateId and status for polling.
• Error handling: responses may indicate insufficient_balance, unauthorized, missing --prompt, or server errors, so the agent should handle recharge, key setup, or retry instead of treating backend failure as a model limitation.

Boundaries

It does not run Midjourney locally or bypass dLazy cloud inference; local media files are uploaded to dLazy media storage when needed by the API. It fits engineering workflows that need a stable image generation endpoint, not scenarios requiring offline generation or strict local-only privacy.