Unsloth AI Documentation

Provides access to Unsloth AI documentation through MCP tools for searching, quickstart, models, tutorials, and installation guidance.

python

0

GitHub Stars

python

Language

4 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "jdanas-unsloth-mcp": {
      "command": "python",
      "args": [
        "unsloth_mcp_server.py"
      ]
    }
  }
}

You can query Unsloth AI documentation through a lightweight MCP server that exposes focused tools for searching topics, obtaining setup guidance, and learning about models and tutorials. This lets MCP clients quickly access relevant documentation content without leaving your development environment.

How to use

Use an MCP client to connect to the server and invoke its tools by name. The available tools let you search the Unsloth documentation, retrieve quickstart instructions, list supported models, fetch tutorials and fine-tuning guides, and obtain installation details.

Run the server in one of these ways, then call the tools from your MCP client. You can start the server directly with Python, use the FastMCP CLI to run it, or test with the provided client.

Direct Python execution to start the server:

python unsloth_mcp_server.py

Using the FastMCP CLI to start the server:

fastmcp run unsloth_mcp_server.py:mcp

Using the test client to verify functionality:

python test_client.py

Example of calling a tool from a client to search documentation:

result = await client.call_tool("search_unsloth_docs", {"query": "fine-tuning"})

How to install

Prerequisites you need on your machine: Python 3.x, and a working Python package manager (pip). You may also use uv to run commands as shown below.

Step-by-step installation flow:

# 1. Clone or download the MCP server package
# (Get the source files you will run locally)

# 2. Create and activate a virtual environment (optional but recommended)
python -m venv venv
source venv/bin/activate  # on Unix/macOS
venv\Scripts\activate     # on Windows

# 3. Install dependencies
pip install -r requirements.txt

# Optional alternative using uvx as shown in examples
uv pip install -r requirements.txt

Configuration and runtime notes

The server is designed to be started and connected via standard MCP workflows. You can run it directly with Python, or via the FastMCP CLI for orchestration. Use a client to call tools by name, supplying the required parameters as shown in tool examples.

Available tools and their purposes include:

Tools exposed by the MCP server

search_unsloth_docs: Search the Unsloth documentation for topics or keywords.
get_unsloth_quickstart: Retrieve the quickstart guide and installation instructions.
get_unsloth_models: Retrieve information about supported models.
get_unsloth_tutorials: Retrieve tutorials and fine-tuning guides.
get_unsloth_installation: Retrieve detailed installation instructions.

Available tools

search_unsloth_docs

Searches the Unsloth documentation for topics or keywords and returns matching sections.

get_unsloth_quickstart

Provides the quickstart guide and basic setup information to get started quickly.

get_unsloth_models

Returns information about models supported by Unsloth.

get_unsloth_tutorials

Returns tutorials and guides, including fine-tuning information.

get_unsloth_installation

Delivers detailed installation instructions for setting up Unsloth-related tooling.