Skip to content
MCP n24q02m logoMCP n24q02m logoMCP n24q02m

Overview

Imagine MCP – Overview

Section titled “Imagine MCP – Overview”

Image and video understanding and generation for AI agents, using your own provider API keys. One server for “look at this media and tell me X” and “make an image or video from this prompt”.

What it does

Section titled “What it does”
  • Understand: answer a prompt about one or more image/video URLs in a single call (mixed image+video where the chosen provider supports it).
  • Generate: create an image, or a video, from a text prompt. Video generation is asynchronous – the first call returns a job id, and you poll with that id until the render completes.
  • Optional reference image, aspect ratio, and duration controls, plus a response cache.

Tools

Section titled “Tools”

Two single-purpose tools – understand and generate – plus the universal config (an action-driven dispatcher covering credentials, runtime settings, and status) and help; config__open_relay reopens the browser credential form. Every parameter is listed in the tools reference.

Clients

Section titled “Clients”

Defaults to local-relay for a single-user browser credential flow and also runs over stdio; self-host the HTTP build for claude.ai web compatibility and multi-user isolation. Works with any MCP client – Claude Code, Codex, Gemini CLI, Cursor, and Windsurf. See the modes overview.

Configuration

Section titled “Configuration”

Bring your own provider API keys (XAI_API_KEY, GEMINI_API_KEY, OPENAI_API_KEY), supplied via the browser relay form or environment variables – with none configured the server reports a degraded state. A default provider and quality tier are runtime settings you can change through the config tool. Use placeholders in shared config and never commit real keys. Full walkthrough in setup.

Next steps

Section titled “Next steps”