generate_imageLABSCO SUMMARY
One call from prompt to a Flux image.
It keeps image generation inside the conversation instead of a browser tab, and because shape, format and quality are arguments on the same call, the picture arrives ready for whatever you were building.
What it is
A one-tool server that sends a prompt to Replicate's Flux 1.1 Pro model and returns the image. The same call carries the aspect ratio, the output format and quality, a safety tolerance, and a switch that has the model expand a terse prompt before rendering.
What you get
- An image generated from a text prompt, rendered by Flux 1.1 Pro
- The aspect ratio set on the call, so the picture arrives at the shape the layout needs
- Output format and quality chosen per call rather than converted afterwards
- Prompt upsampling as a switch, for when the prompt is shorter than the result deserves
- A safety tolerance set on the same call
Requirements
A Replicate account and REPLICATE_API_TOKEN - in the client config, exported in the shell, or read from a .env file. Python 3.10+: clone the repository and pip install -e ., which provides the mcp-replicate-designer command your client runs. The README's install lines, npx route included, point at a placeholder account (github.com/yourusername), so substitute this repository when you copy them.
Setup effort
Build from source — git clone https://github.com/noeltg77/Replicate-Designer.git
USE CASES
01
Getting artwork at the shape the layout needs
Set the aspect ratio on the call - a hero band, a square card, a portrait - so the image does not have to be cropped afterwards.
02
Producing a file the next tool will accept
Choose the output format and quality on the same call, rather than rendering once and converting after.
03
Getting more out of a one-line prompt
Turn prompt upsampling on and let the model expand a terse description before it renders, when you want a look rather than a specification.
04
Holding generated imagery to one standard
Set the safety tolerance on the call so the same bar applies to every image the assistant produces.
1 tool
Listed in project docs. Your client may show a different list.
BEFORE YOU INSTALL
What to know first
Tools you get
1 tool named in the project's documentation.
Requires
- Python 3.10+
- REPLICATE_API_TOKEN in your environment
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
This one is built from source rather than installed with a single command.
Server setup
TERMINAL
git clone https://github.com/noeltg77/Replicate-Designer.git
cd Replicate-Designer
pip install -e .MCP Clients
Any MCP client
ANY MCP CLIENT · JSON
{
"mcpServers": {
"replicateDesigner": {
"command": "mcp-replicate-designer",
"env": {
"REPLICATE_API_TOKEN": "<YOUR_REPLICATE_API_TOKEN>"
}
}
}
}pip install -e . puts the mcp-replicate-designer command on the PATH of the Python environment you ran it in. If your client cannot find it, put that command's full path in "command".
