czg86389-hub/muse2api

What it solves

MUSE2API provides a way to access the advanced multimodal capabilities of muse.ai via a standardized, OpenAI-compatible RESTful API. It solves the problem of needing to manually interact with the muse.ai web interface by reverse-engineering the platform's functionality into an API, allowing for automated text-to-text, text-to-image, image-to-image, and text-to-video workflows.

How it works

The project uses a headless browser (via the CDP protocol) and WebSocket tunnels to interact with the muse.ai platform. It manages multiple accounts in isolated Chromium browser contexts to ensure concurrency and prevent session interference. To maintain access, it includes a background heartbeat system that automatically renews sessions every 48 hours and wakes up cloud VM workspaces. It also provides a Chrome extension and Tampermonkey script to easily export account cookies from a browser to the API server.

Who it’s for

Developers and automation engineers who want to integrate muse.ai's multimodal models (text, image, and video generation) into their own applications or tools using a standard API format.

Highlights

• OpenAI Compatible: Supports /v1/chat/completions, /v1/images/generations, and /v1/images/edits endpoints.
• Multimodal Support: Native support for text dialogue, text-to-image, image-to-image editing, and text/image-to-video generation.
• Account Management: Supports a pool of multiple accounts with intelligent rotation, fault tolerance, and automatic session renewal.
• Councurrency: Uses isolated sandboxes to allow multiple tasks to run in parallel without blocking.
• Operational Tools: Includes a built-in Web Console for health monitoring, account management, and one-click online upgrades.
• Easy Deployment: Ready for Docker Compose or bare-metal installation on Ubuntu/Debian.