One

v1.0.0Developer Toolsstable

A lightweight MCP server that enables intelligent tool management and semantic search for APIs using sentence-transformers. It supports both REST and MCP interfaces across dual transport modes, allowing users to upload, manage, and query API tools wi

ai-agentmcpmcp-convert-to-skillmcp-managermcp-proxy
Share:
392
Stars
0
Downloads
0
Weekly
0/5

What is One?

One is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to lightweight mcp server that enables intelligent tool management and semantic search for apis using sentence-transformers. it supports both rest and mcp interfaces across dual transport modes, allowing...

A lightweight MCP server that enables intelligent tool management and semantic search for APIs using sentence-transformers. It supports both REST and MCP interfaces across dual transport modes, allowing users to upload, manage, and query API tools wi

This server falls under the Developer Tools category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • A lightweight MCP server that enables intelligent tool manag

Use Cases

Manage and search APIs using sentence transformers for semantic matching.
Upload, query, and execute API tools through dual REST and MCP interfaces.
freakynit

Maintainer

LicenseMIT License
Languagego
Versionv1.0.0
UpdatedMay 20, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

Manual Installation

npx one

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use One

one-mcp is a lightweight, self-hosted MCP server written in Go (with a Python implementation) that provides intelligent semantic search over a custom collection of API tool definitions using sentence-transformers embeddings. Developers upload their own API tool schemas in JSON format, then query them by natural language description — the server finds the most relevant tools using vector similarity search. It supports dual transport modes (stdio for MCP clients, HTTP for REST callers) and can act as a dynamic tool registry or proxy hub, letting AI agents discover and invoke the right API tools from a managed library without hardcoding tool lists.

Prerequisites

  • Python 3.9+ with pip (the server is implemented in Python despite being categorized as Go)
  • Git to clone the repository
  • An MCP-compatible client such as Claude Desktop or any MCP-capable tool
  • No API keys required — uses local sentence-transformer embedding models
1

Clone the repository

Download the one-mcp source code to your local machine.

git clone https://github.com/freakynit/one-mcp.git
cd one-mcp
2

Create a virtual environment and install dependencies

Set up a Python virtual environment and install the required packages including sentence-transformers.

python -m venv venv
source venv/bin/activate   # macOS/Linux
# venv\Scripts\activate    # Windows
pip install -r requirements.txt
3

Start the server in stdio mode (for MCP clients)

Launch the server in stdio transport mode, which is the standard mode for MCP client connections.

python main.py --transport stdio
4

Or start with HTTP and stdio combined

Run both transports simultaneously to serve MCP clients via stdio and REST API callers via HTTP on port 8000.

python main.py --transport stdio,http --port 8000
5

Upload API tool definitions

POST your API tool schemas in JSON format to the REST endpoint. Each tool entry should include a name, description, and parameter definitions.

curl -X POST http://localhost:8000/api/tools/upload-json \
  -H "Content-Type: application/json" \
  -d '{"tools": [{"name": "get_weather", "description": "Get weather forecast for a city", "parameters": {}}]}'
6

Configure your MCP client

Add one-mcp to your claude_desktop_config.json pointing at the Python main.py entry point with the virtual environment's Python binary.

{
  "mcpServers": {
    "one-mcp": {
      "command": "/path/to/one-mcp/venv/bin/python",
      "args": ["/path/to/one-mcp/main.py", "--transport", "stdio"]
    }
  }
}

One Examples

Client configuration

Connect Claude Desktop to one-mcp using the venv Python binary to run the stdio server. Update paths to match your installation directory.

{
  "mcpServers": {
    "one-mcp": {
      "command": "/path/to/one-mcp/venv/bin/python",
      "args": ["/path/to/one-mcp/main.py", "--transport", "stdio",
               "--storage_path", "/path/to/one-mcp/tool_embeddings.json"]
    }
  }
}

Prompts to try

After uploading tool definitions, use these prompts to explore semantic tool discovery through the MCP interface.

- "Search for API tools related to weather forecasting"
- "Find tools that can send email notifications"
- "What tools are available for database querying?"
- "Show stats on how many tools are currently registered"
- "Search for tools matching 'user authentication and login'"

Troubleshooting One

Sentence-transformer model download fails or is very slow on first run

The embedding model is downloaded from Hugging Face on first startup. Ensure you have internet access and at least 500 MB of free disk space. If behind a firewall, set the HF_ENDPOINT environment variable to a mirror. The model is cached after the first download.

Semantic search returns poor or irrelevant results

The quality of results depends on the descriptions in your uploaded tool schemas. Write clear, specific descriptions for each tool rather than short generic labels. More descriptive text gives the sentence-transformer model better signals for vector matching.

MCP client cannot connect to one-mcp in stdio mode

Ensure the command in your MCP config uses the full absolute path to the venv Python binary, not the system Python. Run '/path/to/venv/bin/python main.py --transport stdio' directly in a terminal first to confirm it starts without errors.

Frequently Asked Questions about One

What is One?

One is a Model Context Protocol (MCP) server that lightweight mcp server that enables intelligent tool management and semantic search for apis using sentence-transformers. it supports both rest and mcp interfaces across dual transport modes, allowing users to upload, manage, and query api tools wi It connects AI assistants to external tools and data sources through a standardized interface.

How do I install One?

Follow the installation instructions on the One GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.

Which AI clients work with One?

One works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is One free to use?

Yes, One is open source and available under the MIT License license. You can use it freely in both personal and commercial projects.

Browse More Developer Tools MCP Servers

Explore all developer tools servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "one": { "command": "npx", "args": ["-y", "one"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use One?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides