Sleeek MCP Server - Production Ready

What is Sleeek MCP Server - Production Ready?

This MCP server provides context-aware photo assessment for SleeekApp, enabling progressive feedback, angle change detection, and constraint learning across photo attempts. It is designed for developers integrating intelligent photo review into iOS applications.

How to use Sleeek MCP Server - Production Ready?

Deploy to Railway or run locally. Set the OPENAI_API_KEY environment variable. For iOS integration, update the bridge URL in MCPClient.swift to point to your deployed server. Send POST requests to /assess with image data, room type, shoot ID, and current angle.

Key features of Sleeek MCP Server - Production Ready

• Context memory across assessment attempts
• Angle change detection (>30° triggers context reset)
• Progressive acceptance (3 attempts maximum)
• Constraint learning and feedback refinement
• Future agentic capabilities (multi‑step planning, cross‑room optimization)

Use cases of Sleeek MCP Server - Production Ready

• Improve room photo composition with progressive, non‑repetitive feedback
• Automatically detect and handle camera angle changes for fresh assessments
• Integrate context‑aware photo review into the SleeekApp iOS client
• Build a foundation for multi‑step, personalized style adaptation

FAQ from Sleeek MCP Server - Production Ready

How do I deploy this server?

Push the code to a GitHub repository, then create a new Railway project from that repo. Add the OPENAI_API_KEY environment variable in Railway and deploy. Railway provides a public URL.

What does the /assess endpoint expect?

It expects a JSON body with imageBase64 (base64‑encoded image), roomType, shootId, and currentAngle (pitch, yaw, roll). It returns feedback, attempt number, score, and acceptability.

How does the server handle angle changes?

If the camera moves more than 30° from the previous assessment, the server detects that change and resets its context, starting a fresh evaluation from the new angle.

What is progressive acceptance?

The server allows up to 3 assessment attempts per context. After the third attempt, it will accept the photo regardless of score, preventing endless feedback loops.

What are the planned future capabilities?

The architecture supports multi‑step planning, cross‑room optimization, learning from all users, personalized style adaptation, and integration with other tools.