home / mcp / flyto core mcp server

Flyto Core MCP Server

Open-source workflow engine with 300+ atomic modules exposed as MCP tools for AI agents.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "flytohub-flyto-core": {
      "command": "python",
      "args": [
        "-m",
        "core.mcp_server"
      ]
    }
  }
}

Flyto Core is an open-source workflow automation engine designed for AI automation. It exposes hundreds of atomic modules that you can compose into reliable automation pipelines and access them via MCP, enabling AI agents and automation tools to discover, inspect, and execute capabilities directly on your machine or in your environment.

How to use

You connect an MCP client to the Flyto Core MCP server to discover available modules, inspect their parameters, and execute them as part of larger workflows. Your AI agents can list modules, fetch module information, and run specific actions without requiring glue code. Once the MCP server is running, you can start building YAML workflows that chain modules together, retry on failures, and handle errors gracefully.

How to install

Prerequisites you need before installation: Python 3.8 or newer, and Git. You may also choose to install a local package manager like pip, which is included with Python.

pip install flyto-core

If you prefer to build from source, you can clone the repository, install dependencies, and run tests or examples.

git clone https://github.com/flytohub/flyto-core.git
cd flyto-core
pip install -r requirements.txt
python run.py workflows/_test/test_text_reverse.yaml

Additional setup for running the MCP server locally

Flyto Core can expose an MCP server that your MCP clients connect to for discovering and executing modules. You start the MCP server with a Python module invocation that runs the MCP server entry point.

python -m core.mcp_server

Available tools

string.reverse

Reverse the characters in a string and return the result.

string.uppercase

Convert a string to uppercase.

browser.launch

Launch a browser instance for automation tasks.

browser.goto

Navigate the browser to a specified URL.

browser.extract

Extract information from a page using a selector.

http.request

Perform HTTP requests with built-in retry and error handling.

ai.ollama.chat

Chat with a local Ollama model for AI-assisted workflows.

model

Load and run AI models within a workflow context.

memory

Store and retrieve memory for agents during task execution.

memory_vector

Handle vector-based memory for semantic search and retrieval.