Embeddr CLI

A local CLI and web UI for searching images and prompts using embeddings.

Installation

Using uv

# Create a Virtual Environment
mkdir embeddr && cd embeddr
uv venv && source .venv/bin/activate

# Install Torch - https://pytorch.org/get-started/locally/

## EXAMPLES
# - CPU ONLY
uv pip install torch torchvision --index-url https://download.pytorch.org/whl/cpu
# OR
# - CUDA 13.0
uv pip install torch torchvision --index-url https://download.pytorch.org/whl/cu130

# Install Embeddr
uv pip install embeddr-cli

Usage

Once installed to start Embeddr

> embeddr serve

✨ Embeddr Local API has started!
   ---------------------------------------------
   👉 Web UI:    http://127.0.0.1:8003
   ---------------------------------------------
   Press Ctrl+C to stop server

embeddr serve

--help   Shows help.
--mcp    Enables Model Context Protocol.
--docs   Enable API docs.
--host   The host to bind to. [default: 127.0.0.1]
--port   The port to bind to. [default: 8003]

embeddr config

--init   Shows help.
--show   Shows the current configuration.

Integrations

ComfyUI Extension

comfy node install embeddr-extension

View Registry

Mistral Vibe MCP

• Run embeddr with embeddr serve --mcp
• Add this to your .vibe/config.toml
• vibe

[[mcp_servers]]
name = "embeddr"
transport = "http"
url = "http://localhost:8003/mcp/messages"

Anything that uses Model Context Protocol will also work.

Screenshots

Home Page

Settings Page

Development

Instructions to run a development version.

git clone https://github.com/embeddr-net/embeddr-cli
cd embeddr-cli
uv venv
uv pip install -e .

# Install Torch

# Start Embeddr API with live reload
uv run embeddr serve --reload 

While releases do include the frontend, this repo does not.

You can run a development frontend or download the latest version and extract it into embeddr-cli/src/web/

Packages