# MeltFlex AI [Health: Active]

**Category:** 📐 Architecture & Design  
**Repository:** https://github.com/MeltFlexDevs/skills  
**GitHub Stars:** 1  
**npm Downloads (last month):** 414  
**Views:** 4  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/meltflex-ai

## Description
AI interior design — restyle rooms, place furniture, and virtually stage spaces from a photo.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "meltflex-ai": {
    "command": "npx",
    "args": ["-y","meltflex-mcp"],
    "env": {
      "MELTFLEX_API_KEY": ""
    }
  }
}
```

**Requires environment variables:** `MELTFLEX_API_KEY` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What MeltFlex AI does

The MeltFlex AI MCP server connects an AI agent to MeltFlex's image, video, and 3D generation API. It is aimed at architecture and design workflows involving rooms, buildings, gardens, furniture, floorplans, and 3D visualization.

The design skill supports 16 image modes, including room restyling, virtual staging, layout improvement, decluttering, exterior and garden design, wall and floor changes, stairs, doors, windows, kitchen and bathroom redesigns, photo rendering from a 3D draft, seasonal variations, and masked region edits. A separate furniture skill places specified real products into a room photo using up to 10 reference images. The 3D skill covers furniture photos converted to models, floorplans converted to models or rendered images, and room photos converted to explorable 3D worlds.

## How it works

The skills call MeltFlex's public API directly. A source photo, floorplan, or reference image is sent with the selected operation and optional generation settings. Image requests can specify 512, 1K, or 2K resolution, a design level, one to three variations, and reference image URLs. Region editing uses a mask and changes one marked area.

The repository exposes Markdown-based skills such as `/meltflex:design`, `/meltflex:furniture`, and `/meltflex:3d`. It also supports MCP-speaking clients and agents that can load these skills. Longer 3D jobs can return HTTP 202 with a polling URL; the 3D skill handles that polling flow. Listed MCP operations include video generation, floorplan-to-3D conversion, furniture-to-3D conversion, and world generation. The floorplan rendered-picture option is documented as REST-only.

## Setup and configuration

Install the repository with `npx skills add MeltFlexDevs/skills`, GitHub CLI, the Claude Code marketplace, or the included setup script. The project also documents installation paths for Cline and other MCP-settings agents.

An active paid MeltFlex subscription is required. Generate an API key from MeltFlex account settings and expose it as `MELTFLEX_API_KEY`. Alternatively, the MCP or CLI login command opens MeltFlex's browser authorization flow and stores the resulting key in `~/.meltflex/config.json` with restricted file permissions.

The MeltFlex AI MCP server bills every generation to the authenticated user's account. Image generations cost 10 credits each, with failed images automatically refunded. Other documented costs include 100 or 150 credits for a 4- or 8-second video, 75 credits for a furniture model, 10 or 100 credits for floorplan output, and 30 or 200 credits for draft or HD worlds.

## Tools and capabilities

- Restyle rooms, exteriors, gardens, kitchens, and bathrooms from photos.
- Apply floor, wall, stair, door, or window changes.
- Furnish spaces with specified products and reference images.
- Produce cinematic walkthrough videos in MP4 format.
- Convert furniture photos to GLB models, with additional USDZ, FBX, and OBJ output noted for that operation.
- Convert floorplans to rendered 3D pictures or GLB models, and create explorable worlds from room photos.

## Limitations and notes

MeltFlex generation requires an active paid subscription, an API key, and available account credits. The documented 3D operations can take about one to five minutes, while video generation takes about one minute. Results that outlive the request use asynchronous polling.

The repository does not describe a local generation runtime; its skills call the public MeltFlex API. It also distinguishes between MCP-supported operations and the floorplan-to-render REST option. Treat the published credit amounts and processing times as operation-specific reference values rather than a single fixed cost or response time.

For developers evaluating the MeltFlex AI MCP server in the Architecture & Design category, the main fit is agent-assisted visual iteration rather than general-purpose 3D authoring or CAD editing. The provided material documents image, video, model, and world generation, but does not describe direct editing of generated 3D assets.

## Getting started with this MeltFlex AI MCP server
Always refer to the official documentation for the most accurate and up-to-date information.

_Full upstream README: https://allmcps.com/mcp/meltflex-ai/readme_