Bespoke Automata

Bespoke Automata, An Introduction

About Bespoke Automata

Create and deploy sophisticated Agent AI's to a single API with Bespoke Automata. With Bespoke Automata, you can combine large language models running locally or remotely with instruments for database IO, dictionaries, arrays, logic, APIs and more into powerful Brains capable of pursuing goals set by their designers.

With Bespoke Automata, you can design and test brains via a Directed Graph GUI (powered by litegraph), and deploy them behind a single user friendly API, each brain a different endpoint. Brains are exposed as Model Context Protocol (MCP) tools, making them accessible to any MCP-compatible AI assistant.

Demo Video

Installation

Requirements

• Node.js (v18+)
• NPM
• Python (for the optional LLM API)
  • flask
  • sentence_transformers

Quick Start

Clone the repository and install dependencies:

$ git clone https://github.com/C0deMunk33/bespoke_automata
$ cd bespoke_automata
$ npm install
$ npm start

Open your browser to http://localhost:9999. The app has three tabs:

• Editor — visual node graph editor (powered by LiteGraph) for designing brains
• Manager — view deployed brains, their schemas, and MCP connection info
• Runner — execute brains with custom inputs and see outputs

Brains are saved to the graphs/ directory. Saving in the editor makes them immediately available to the MCP server and runner.

MCP Integration

The server exposes an MCP SSE endpoint at http://localhost:9999/sse. Each brain in graphs/ is registered as an MCP tool. Connect any MCP-compatible client (Cursor, Claude Desktop, etc.) to this endpoint to use your brains as AI tools.

Configuration

API Endpoints

Standalone MCP Server

To run only the MCP server without the web UI:

$ npm run mcp

BA API (Optional LLM Backend)

The BA API uses llama-cpp-python for local text inference and vision.

• Place text models in ../models/text
• Place vision models in ../models/vision
• NOTE: Model directories are at the same level as this repo. GGUF format works best — get them from Hugging Face.

Installation by platform:

• Metal (macOS): CMAKE_ARGS="-DLLAMA_METAL=on" pip install llama-cpp-python
• CUDA (Linux): CMAKE_ARGS="-DLLAMA_CUBLAS=on" pip install llama-cpp-python
• CUDA (Windows): $env:CMAKE_ARGS = "-DLLAMA_CUBLAS=on" then pip install llama-cpp-python
• CPU: pip install llama-cpp-python

Then run:

$ cd APIs/
$ python omni_api.py

The API server will be at your_ip:5000.

NOTE: On macOS, port 5000 collides with Airplay Receiver. Turn it off in Settings > General > Airdrop & Handoff, or change the port in the config.

Project Structure

bespoke_automata/
├── server.js                 # Unified Express server
├── package.json              # Dependencies
├── graphs/                   # Brain files (.brain)
├── libs/                     # LiteGraph, nodes, and supporting libraries
├── public/                   # Web UI (SPA)
│   └── index.html
├── bespoke_manager/
│   ├── core.js               # Brain execution engine
│   ├── mcp_server.js         # Standalone MCP server
│   ├── run.js                # CLI brain runner
│   └── front_ends/           # Discord bot integrations
├── APIs/                     # Python LLM API servers
└── README.md

More Info

• Demo Brains: ./graphs/
• Example: https://youtu.be/w_saaTFEuSM
• Issues: Create an issue here or ping me at: https://twitter.com/icodeagents
• Contact: https://twitter.com/icodeagents
• Discord: https://discord.gg/Rr5Ac5GFeY

THANKS AND GOOD LUCK!!