Chess UCI
About
Connect to UCI-compatible chess engines like Stockfish to play and analyze games. Requires a local chess engine binary.
Details
- Author
- AnglerfishChess
- GitHub stars
- 1
- Downloads
- 359
- Categories
- Other, AI
Jump to
- Bridges MCP with any UCI-compatible chess engine
- Supports configurable UCI options (e.g., Threads, Hash)
- Provides chess analysis via FEN strings
- Offers best‑move retrieval for a given position
- Allows setting the current board position explicitly
- Reports engine details and supports debug logging
Setting up with Highlight
This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:
- Download and install Highlight from highlightai.com/download
- Navigate to the plugins tab and select "Add Custom Plugin"
-
Configure the plugin with the settings below
Plugin Name
Chess UCICommand (node, npx, python, etc.)Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.
- Enable "Start Automatically" if you want the plugin to start when Highlight launches
From the repository
Requires Python 3.10 or newer, uv/uvx, and a UCI-compatible chess engine binary (e.g., Stockfish). Configure it in your MCP client settings (e.g., claude_desktop_config.json) using either uvx (recommended) or uv commands, specifying the engine path and optional UCI options with -o. Run the server by launching the client; commands are invoked via MCP tools.
Claude Desktop / Cursor
Paste into your MCP client config file to install this server.
{
"mcpServers": {
"chess uci": {
"chess-uci-mcp": {
"command": "uv",
"args": [
"venv",
"--python",
"python3.10"
]
}
}
}
}McpServers
{
"chess-uci-mcp": {
"command": "uv",
"args": [
"venv",
"--python",
"python3.10"
]
}
}An MCP bridge that provides an interface to UCI chess engines (such as Stockfish or Leela Chess Zero).
You need to have Python 3.10 or newer, and alsouv/uvxinstalled.
To function, it requires an installed UCI-compatible chess engine, like Stockfish (has been tested with Stockfish 17).
In case of Stockfish, you can download it fromhttps://stockfishchess.org/download/.
On macOS, you can usebrew install stockfish.
You need to find out the path to your UCI-capable engine binary; for further example configuration, the path is e.g./usr/local/bin/stockfish(which is default for Stockfish installed on macOS using Brew).
The further configuration should be done in your MCP setup; for Claude Desktop, this is the fileclaude_desktop_config.json(find it inSettingsmenu,Developer, thenEdit Config).
- macOS:~/Library/Application\ Support/Claude/claude_desktop_config.json
- Windows:%APPDATA%/Claude/claude_desktop_config.json
- Linux:~/.config/Claude/claude_desktop_config.json
Add the following settings to your MCP configuration (depending on the way to run it you prefer):
Uvx is able to directly run the Python application by its name, ensuring all the dependencies, in a automatically-created virtual environment. This is the preferred way to run thechess-uci-mcpbridge.
Set up your MCP server configuration (e.g. Claude Desktop configuration) file as following:
"mcpServers": { "chess-uci-mcp": { "command": "uvx", "args": ["chess-uci-mcp@latest", "/usr/local/bin/stockfish"] } }
To pass options to the engine, add them to theargsarray. For example, to set theThreadsandHashoptions for Stockfish:
"mcpServers": { "chess-uci-mcp": { "command": "uvx", "args": [ "chess-uci-mcp@latest", "/usr/local/bin/stockfish", "-o", "Threads", "4", "-o", "Hash", "128" ] } }
Use it if you have the repository cloned locally and run from it:
"mcpServers": { "chess-uci-mcp": { "command": "uv", "args": ["run", "chess-uci-mcp", "/usr/local/bin/stockfish"] } }
Similarly, to pass options when running withuv:
"mcpServers": { "chess-uci-mcp": { "command": "uv", "args": [ "run", "chess-uci-mcp", "/usr/local/bin/stockfish", "-o", "Threads", "4", "-o", "Hash", "128" ] } }
The application accepts the following command-line options:
- ENGINE_PATH: (Required) The path to the UCI-compatible chess engine executable.
- --uci-optionor-o: Set a UCI option. This option can be used multiple times. It takes two arguments: the option name and its value (e.g.,-o Threads 4).
- --think-time: The default thinking time for the engine in milliseconds. Defaults to1000.
- --debug: Enable debug logging.
The bridge provides the following MCP commands:
- analyze- Analyze a chess position specified by FEN string
- get_best_move- Get the best move for a chess position
- set_position- Set the current chess position
- engine_info- Get information about the chess engine
- get_engine_options- Get all available UCI engine options with their metadata and current values
- set_engine_options- Set one or more UCI engine options at runtime
# Clone the repository git clone https://github.com/AnglerfishChess/chess-uci-mcp.git # ... or # git clone git@github.com:AnglerfishChess/chess-uci-mcp.git cd chess-uci-mcp # Create a virtual environment uv venv --python python3.10 # Activate the virtual environment source .venv/bin/activate # On Unix/macOS # or .venv\Scripts\activate # On Windows # Install the package in development mode # uv pip install -e . # or, with development dependencies uv pip install -e ".[dev]" # Resync the packages: uv sync --extra=dev # Run tests pytest # Check code style ruff check
The checklist lives in thereleasingskill under.claude/skills/, so a release runs the same way every time: preconditions, version bump, tag, GitHub release. Publishing a GitHub release is the trigger — from there.github/workflows/publish.ymlbuilds the package and uploads it to PyPI through atrusted publisher, then republishes the MCP registry entry. Both authenticate over OIDC, so no token is stored in this repository or on any developer's machine.
Nothing is automatic: a release only happens when a human publishes the GitHub release.
pyproject.tomlholds the version, and every other copy is derived from it:
uv sync --extra=dev # updates uv.lock, keeping the dev tools installed uv run python .claude/skills/releasing/scripts/sync_version.py
That writeschess_uci_mcp/__init__.pyand both version fields inserver.json. Passing--checkinstead reports drift without touching anything, which is what CI runs.
curl -s "https://registry.modelcontextprotocol.io/v0/servers?search=chess-uci-mcp&limit=3"
The listing is described byserver.json, under the nameio.github.AnglerfishChess/chess-uci-mcp. GitHub authentication grants theio.github.<user>/namespace; an organisation namespace additionally requires Owner rights on that organisation, and the name is case-sensitive.
Ownership of the PyPI package is proven by themcp-name:marker near the top of this README, which becomes the package description on PyPI. The registry reads it from thepublished*artifact, so adding it to git is not enough — it only counts once a release carrying it reaches PyPI. Note also that the registry capsdescriptionat 100 characters where PyPI does not, which is whyserver.jsoncarries its own one-line description rather than reusing the project's.
Live AI agent city-economy on Base L2 — 92 NPC agents earn real USDC via jobs, trading, and businesses across 10 cities. Claim jobs for USDC payouts, chat with agents, check AGWC token data, and get live crypto news. 11 tools, no API key.
An MCP server for '海龟汤' (Turtle Soup), a scenario-based reasoning puzzle game.
Play a legendary text adventure by talking to your AI — no commands to memorize. The Hidden Empire puts a full underground world of puzzles, treasures, and trolls inside your conversation. Speak naturally: say 'head north,' 'grab the lantern,' or 'what am I carrying?' and your AI handles the rest. Execute multi-move plans in one shot, undo mistakes instantly, and save up to 20 named playthroughs you can resume from any session. Based on the MIT-licensed Zork I source, rebuilt from the ground up for AI-native play.
An AI-powered, synthwave, maze-crawling, and world-building adventure game server.
Control a Minecraft character to build, explore, and interact with the game world using natural language.
Enables AI assistants to interact with a Minecraft server using the Model Context Protocol (MCP).
A massively multiplayer online game for AI agents -- pilot spaceships, mine, trade, craft, explore, and battle in a galaxy of ~500 systems via MCP.
Play a roguelike MMO as an AI agent. Explore, fight, chat, and survive tsunamis via 50+ MCP tools
MCP server for Turtle Soup (lateral thinking puzzles). Start sessions, ask questions, get 4-class judgments (Yes/No/Both/Irrelevant), and reveal the full story when allowed.
Headless Space MMO where you are the CEO and your agent is your staff
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.
