mcp -demo

This MCP server exposes four tools built on top of FastMCP: create_visualization, text_to_speech, analyze_image, and create_video. After initializing a session, you can call these tools to generate visual content, synthesize speech, analyze images, or produce videos from text prompts or media. To begin, initialize a session to obtain an mcp-session-id, then use the tools/endpoints with that session ID in subsequent requests. The HTTP transport workflow mirrors standard MCP interactions: initialize, receive a session ID, then call tools with the appropriate parameters.

Tool capabilities:

• create_visualization: Generates an image from a text prompt with optional parameters such as aspect_ratio. Example usage involves sending a request with name: create_visualization and arguments including prompt and aspect_ratio.
• text_to_speech: Converts text into speech audio. Provide a prompt (text) and choose a voice_name if desired.
• analyze_image: Analyzes an image from a URL or embedded input and returns extracted details such as text or objects detected, guided by a prompt that describes what to extract.
• create_video: Creates a video from a textual prompt or a set of images. You can supply a prompt and an optional filename_hint for storage.

Example flow: initialize to get mcp-session-id, then call tools/call with name set to one of the tool names and pass the corresponding arguments. Ensure you pass your API key via x-api-key and the session ID in mcp-session-id for each operation.

Prerequisites:

• Python 3.11+
• uv (Python package manager/runner)
• Access to the repository (git)
• A valid Gemini API key (GEMINI_API_KEY)

1. Clone the repository and install dependencies

git clone https://github.com/Innovate-Lab/mcp-server-demo
cd MCP
uv sync

2. Configure environment

cp .env.example .env

Edit the .env file if needed to set BASE_URL, GEMINI_API_KEY, and optional GCS_BUCKET.

3) Run the server
```bash
uv run python -m src.main
# Server: http://localhost:8000 | Health: /health

Environment variables and configuration tips:

• GEMINI_API_KEY must be set to your Google Cloud API key for Gemini-powered capabilities.
• BASE_URL should reflect where the server is reachable (default http://localhost:8000).
• GCS_BUCKET is optional; if set, uploads will go to Google Cloud Storage and files will be served from there. If not set, files are stored locally under ./static/.
• Health endpoint is available at /health; the main API is at /mcp.
• If you modify ports or base paths, ensure all clients use the updated BASE_URL.
• The project structure keeps source logic under src/, with tools handling image, audio, and video processing, and storage.py managing uploads.
• For local testing, ensure network access to any external services required by the tools (e.g., Gemini API).