wtf is midjourney-proxy?

midjourney-proxy lets you use MidJourney's image generation through a standard API instead of typing commands in Discord. Normally, you create images by sending prompts into a Discord channel and waiting for results. This tool acts as a middle layer so your own apps can request images programmatically and get them back through a web interface your code can talk to. To set it up, you create your own Discord server with a channel for MidJourney, then grab your user token, server ID, and channel ID from Discord. The proxy connects to that channel on your behalf and exposes a set of API endpoints. You can send an Imagine prompt with an optional reference image, ask it to describe an image and generate a prompt, blend multiple images together, and track generation progress. It also translates Chinese prompts and filters out banned words. Tasks go through a queue (10 queued, 3 concurrent by default) so requests don't overwhelm your MidJourney account. Someone building an AI-powered design tool, a chatbot that generates images, or a mobile app for creating artwork would use this to put MidJourney behind their own product. The README lists a few example projects built on top of it, including a WeChat client and a web service combining ChatGPT with MidJourney. The main tradeoff is account risk. Using your personal Discord token to connect gives you full functionality and error messages, but the README warns this may increase the chance of your account being flagged. You can switch to a bot token instead, though that limits some features. Heavy image generation can also trigger MidJourney's verification checks, which require manual intervention. Deployment is available through Railway (no server needed) or Docker, and local development requires Java 17 and Maven.