dotty-mcp 0.1.0

MCP server for Scala 3 (Dotty) compiler development

dotty-mcp

A Model Context Protocol (MCP) server for Scala 3 (Dotty) compiler development.

Overview

dotty-mcp provides a streamlined interface for interacting with the Scala 3 compiler through SBT. It maintains a persistent SBT session and exposes compiler functionality through the MCP protocol, making it ideal for AI-assisted compiler development workflows.

Features

• Persistent SBT Session: Maintains a long-running SBT process for fast compilation
• Direct scalac Access: Compile individual files with custom compiler options
• MCP Integration: Works seamlessly with Claude Code and other MCP clients

Installation

For Claude Code

From a local clone:

cd /path/to/dotty-mcp
uv sync
claude mcp add dotty-mcp -- uv run --directory /path/to/dotty-mcp dotty-mcp --root /path/to/dotty-project

After publishing to PyPI:

# If running from within your Dotty project directory
claude mcp add dotty-mcp -- uvx dotty-mcp

# Or specify a custom root
claude mcp add dotty-mcp -- uvx dotty-mcp --root /path/to/dotty-project

Manual Installation

# Install dependencies
uv sync

# Run the server (from within Dotty project directory)
uv run dotty-mcp

# Or specify a custom root
uv run dotty-mcp --root /path/to/dotty-project

Note: The --root argument is optional and defaults to the current directory.

Usage

The server provides a single tool:

scalac(file: str, options: List[str] = None) -> str

Compiles a Scala file using the Dotty compiler through SBT.

Parameters:

• file: Relative path from project root to the Scala file (e.g., "tests/pos/HelloWorld.scala")
• options: Optional list of compiler options (e.g., ["-Xprint:typer", "-explain"])

Returns: Compilation output including errors, warnings, or success messages.

Example:

scalac("tests/pos/Test.scala", ["-Xprint:typer"])

Requirements

• Python >= 3.10
• SBT (must be in PATH)
• A Dotty/Scala 3 project with a build.sbt file

Architecture

The tool consists of three main components:

1. SBTProcess: Manages the persistent SBT process using pexpect
2. DottyProject: Wraps the SBT process and provides high-level compilation operations
3. MCP Server: Exposes functionality through the FastMCP framework

Development

This project uses uv for dependency management:

# Install dependencies
uv sync

# Run tests (if available)
uv run pytest

License

MIT